Version: dev

NoahSoilMoisture

Soil water transport of WRF's unified Noah land-surface model as instantaneous tendencies: SUBROUTINES SMFLX, SRT and WDFCND of phys/module_sf_noahlsm.F (WRF 4.8, sf_surface_physics = 2), together with the canopy-interception budget and the three runoff terms. THE PDE is Richards' equation in mixed diffusivity/conductivity form, d(theta)/dt = d/dz( D(theta) d(theta)/dz - K(theta) ) - S, with Clapp-Hornberger closures D = dwsat (theta/smcmax)^(bexp+2) and K = dksat (theta/smcmax)^(2 bexp+3). It is written as the flux-form operator D(K_w D(sh2o) - F_w) and lowered by the SAME EarthSciDiscretizations rule the soil heat equation uses, column_nonuniform_1d/rules/varcoeff_face_flux_laplacian_lev -- NO NEW ESD RULE WAS NEEDED. That works because WRF evaluates the gravity drainage K(theta) UPWIND, at the layer above each interface, so from the operator's point of view it is a PRESCRIBED interface flux and drops straight into the rule's F slot; a scheme that averaged K across the face would not fit and would have needed a new rule. The same upwinding applies to the diffusivity (FORTRAN_BUGS N90, as in the heat equation). Orientation and geometry follow soil_heat.esm: `lev` is the soil layer index with layer 1 at the surface and `ze` the interface DEPTH below ground, positive downward. BOUNDARY FLUXES. The surface interface carries the infiltration pddum, and the bottom interface the free drainage slope wcnd[NLEV], which IS the subsurface runoff; both are prescribed, so the diffusive part at the two domain faces is correctly dropped. SINKS are the bare-soil evaporation, which withdraws from layer 1 only, and the per-layer transpiration. THE INFILTRATION LIMITER has a clean dt -> 0 limit that is not obvious from the Fortran: WRF forms infmax = px (ddt/(px + ddt))/dt with px = pcpdrp dt and ddt = dd (1 - exp(-kdt dt/86400)), and BOTH numerator terms are proportional to dt, so the limit is the harmonic-style combination pcpdrp r/(pcpdrp + r) with r = dd kdt/86400 a rate. dd is the column-integrated storage deficit, written here as an integral. The limiter is then floored at the surface conductivity and capped at the supply, and whatever the soil cannot take becomes surface runoff. DEW IS ROUTED LIKE RAIN: where the potential evaporation is negative the surface gains water, and NOPAC adds that gain to the precipitation before the canopy/throughfall split, so a dew step has a small positive canopy tendency and a small positive infiltration even with no precipitation at all. THE CANOPY is a reservoir of capacity cmcmax fed by shdfac prcp and drained by ec; what it cannot hold becomes drip, which joins the throughfall. In the dt -> 0 limit WRF's drip = max(cmc + dt rhsct - cmcmax, 0) becomes a SWITCH: zero while the canopy has room, and the whole net input once it is full, at which point dcmc_dt is zero. The reference set contains a step with the canopy exactly full and raining, so that branch is exercised. NOT TRANSCRIBED, and a consumer must not mount this where they matter: (a) SSTEP's supersaturation redistribution, which cascades water upward out of any layer driven past smcmax and reports the leftover as runoff3 -- that is a PROJECTION of the state applied after an integration step, not a rate, exactly like the slab scheme's snow cap, so it is reported as the indicator supersaturation_active instead; (b) the SMFLX predictor-corrector, which runs SRT/SSTEP twice on a half-updated column when the rainfall is heavy or the soil near saturation -- it is a finite-dt integration device and both branches converge to the SAME right-hand side as dt -> 0, so it does not affect this component, but it also means the corrector path is untested; (c) the frozen-ground infiltration reduction fcr, which IS written here because it is a clean closed form, but which the reference set never exercises (no soil ice anywhere), so it is UNTESTED rather than merely unused, as is every appearance of sice; (d) the WRF_HYDRO surface-head coupling. Consumer-supplied column fields (ze, sh2o, smc, et) are rewrite-target ops `input_<name>` lowered by the consumer, in the inline tests through per-test expression_template_imports libraries generated by tools/fill_tests.py.

Domain
land_surface
Subdomain
noah
Type
model
.esm schema
1.2.0
Source
components/land_surface/noah/soil_moisture.esm
Tags
#land_surface #noah #model #wrf #land-surface #soil-moisture #richards-equation #runoff #column #pde

Description

Soil water transport of WRF’s unified Noah land-surface model as instantaneous tendencies: SUBROUTINES SMFLX, SRT and WDFCND of phys/module_sf_noahlsm.F (WRF 4.8, sf_surface_physics = 2), together with the canopy-interception budget and the three runoff terms. THE PDE is Richards’ equation in mixed diffusivity/conductivity form, d(theta)/dt = d/dz( D(theta) d(theta)/dz - K(theta) ) - S, with Clapp-Hornberger closures D = dwsat (theta/smcmax)^(bexp+2) and K = dksat (theta/smcmax)^(2 bexp+3). It is written as the flux-form operator D(K_w D(sh2o) - F_w) and lowered by the SAME EarthSciDiscretizations rule the soil heat equation uses, column_nonuniform_1d/rules/varcoeff_face_flux_laplacian_lev – NO NEW ESD RULE WAS NEEDED. That works because WRF evaluates the gravity drainage K(theta) UPWIND, at the layer above each interface, so from the operator’s point of view it is a PRESCRIBED interface flux and drops straight into the rule’s F slot; a scheme that averaged K across the face would not fit and would have needed a new rule. The same upwinding applies to the diffusivity (FORTRAN_BUGS N90, as in the heat equation). Orientation and geometry follow soil_heat.esm: lev is the soil layer index with layer 1 at the surface and ze the interface DEPTH below ground, positive downward. BOUNDARY FLUXES. The surface interface carries the infiltration pddum, and the bottom interface the free drainage slope wcnd[NLEV], which IS the subsurface runoff; both are prescribed, so the diffusive part at the two domain faces is correctly dropped. SINKS are the bare-soil evaporation, which withdraws from layer 1 only, and the per-layer transpiration. THE INFILTRATION LIMITER has a clean dt -> 0 limit that is not obvious from the Fortran: WRF forms infmax = px (ddt/(px + ddt))/dt with px = pcpdrp dt and ddt = dd (1 - exp(-kdt dt/86400)), and BOTH numerator terms are proportional to dt, so the limit is the harmonic-style combination pcpdrp r/(pcpdrp + r) with r = dd kdt/86400 a rate. dd is the column-integrated storage deficit, written here as an integral. The limiter is then floored at the surface conductivity and capped at the supply, and whatever the soil cannot take becomes surface runoff. DEW IS ROUTED LIKE RAIN: where the potential evaporation is negative the surface gains water, and NOPAC adds that gain to the precipitation before the canopy/throughfall split, so a dew step has a small positive canopy tendency and a small positive infiltration even with no precipitation at all. THE CANOPY is a reservoir of capacity cmcmax fed by shdfac prcp and drained by ec; what it cannot hold becomes drip, which joins the throughfall. In the dt -> 0 limit WRF’s drip = max(cmc + dt rhsct - cmcmax, 0) becomes a SWITCH: zero while the canopy has room, and the whole net input once it is full, at which point dcmc_dt is zero. The reference set contains a step with the canopy exactly full and raining, so that branch is exercised. NOT TRANSCRIBED, and a consumer must not mount this where they matter: (a) SSTEP’s supersaturation redistribution, which cascades water upward out of any layer driven past smcmax and reports the leftover as runoff3 – that is a PROJECTION of the state applied after an integration step, not a rate, exactly like the slab scheme’s snow cap, so it is reported as the indicator supersaturation_active instead; (b) the SMFLX predictor-corrector, which runs SRT/SSTEP twice on a half-updated column when the rainfall is heavy or the soil near saturation – it is a finite-dt integration device and both branches converge to the SAME right-hand side as dt -> 0, so it does not affect this component, but it also means the corrector path is untested; (c) the frozen-ground infiltration reduction fcr, which IS written here because it is a clean closed form, but which the reference set never exercises (no soil ice anywhere), so it is UNTESTED rather than merely unused, as is every appearance of sice; (d) the WRF_HYDRO surface-head coupling. Consumer-supplied column fields (ze, sh2o, smc, et) are rewrite-target ops input_<name> lowered by the consumer, in the inline tests through per-test expression_template_imports libraries generated by tools/fill_tests.py.

Reference

  • Clapp, R. B., and G. M. Hornberger (1978). Empirical equations for some soil hydraulic properties. Water Resour. Res., 14, 601-604.
  • Schaake, J. C., et al. (1996). Simple water balance model for estimating runoff at different spatial and temporal scales. J. Geophys. Res., 101(D3), 7461-7475.
  • Koren, V., et al. (1999). A parameterization of snowpack and frozen ground intended for NCEP weather and climate models. J. Geophys. Res., 104(D16), 19569-19585.
  • Chen, F., and J. Dudhia (2001). Coupling an advanced land surface-hydrology model with the Penn State-NCAR MM5 modeling system. Part I. Mon. Wea. Rev., 129, 569-585.
  • wrf-model/WRF v4.8.0 phys/module_sf_noahlsm.F SUBROUTINE SMFLX lines 2669-2813, SRT lines 3653-3947, SSTEP lines 3950-4077, WDFCND lines 4461-4519 (instrumented in ctessum-claude/WRF branch earthsciml-instrumented-chem170, commit 342156a3).

Parameters

NameUnitsDefaultDescription
sh2o_inm^3/m^30.0Coupling target (assembly input path): a shaped parameter that receives the unfrozen soil water from an assembly’s ODE state through a document-level variable_map (param_to_var), when that document’s mount-edge library lowers input_sh2o to it (esm-spec 9.7.10, 10.4; couplings/noah_column.esm). Unused by this component’s own inline tests, which lower input_sh2o to dumped constants.
smc_inm^3/m^30.0Coupling target for the total soil moisture, as sh2o_in (lowered from input_smc by an assembly’s mount-edge library).
et_inW/m^20.0Coupling target for the per-layer transpiration, which an assembly hands over from the evapotranspiration component’s et through a document-level variable_map (param_to_var) when its mount-edge library lowers input_et to it.
edirW/m^230.0Direct soil evaporation in energy units, which withdraws from the TOP LAYER ONLY.
ecW/m^250.0Canopy-water evaporation in energy units, which drains the canopy reservoir and never touches the soil.
prcpkg/(m^2*s)0.0Precipitation rate reaching the surface, WRF’s PRCP.
shdfac10.5Green vegetation fraction, which splits the precipitation between canopy interception (shdfac) and direct throughfall (1 - shdfac).
cmcm0.0005Canopy-intercepted water content, WRF’s CMC.
smcmaxm^3/m^30.476Soil porosity, SOILPARM.TBL MAXSMC.
smcwltm^3/m^30.084Soil wilting point, SOILPARM.TBL WLTSMC, which sets the bottom of the storage deficit the infiltration limiter measures.
bexp15.33Clapp-Hornberger b, SOILPARM.TBL BB.
dksatm/s2.81e-06Saturated hydraulic conductivity, SOILPARM.TBL SATDK.
dwsatm^2/s2.39e-05Saturated soil water diffusivity, SOILPARM.TBL SATDW.
kdt14.215Infiltration parameter, refkdt dksat/refdk from REDPRM.
frzx10.1748Per-soil-type frozen-ground parameter, frzk frzfact from REDPRM. Read only by the fcr reduction, which no state in the reference set reaches.
slope10.1Bottom drainage-slope factor, GENPARM.TBL SLOPE_DATA: the fraction of the deepest layer’s gravity drainage that actually leaves the column as subsurface runoff.
etpW/m^2100.0Potential evaporation in energy units from the Penman component. Read ONLY for its sign: where it is negative the surface is gaining dew, and NOPAC adds that gain to the precipitation reaching the canopy and the soil.
meltm/s0.0Snowmelt water arriving at the surface, WRF’s EX, which SNOPAC adds to PRCP1 before calling SMFLX. Zero on a snow-free column, where SFLX takes the NOPAC branch and there is no pack to melt. It joins the supply at exactly the same point as rain and therefore goes through the same canopy split; it is NOT a separate soil source. On a melting pack it dominates everything else: 8.0e-7 m/s at the strongest step of the cold-season set, against a bare-ground column’s dew of order 1e-9.
dew_to_soil11.0Whether dew and frost are routed into the SOIL water, 1 on a snow-free column and 0 under a snow pack. This is a real structural difference between WRF’s two branches and not a switch of convenience: NOPAC does PRCP1 = PRCP1 + DEW, so frost on bare ground infiltrates, while SNOPAC sets DEW = -ETP1 and never adds it to PRCP1 – the frost is deposited on the PACK and reaches the soil only later, as melt. Defaulting to 1 keeps every snow-free test unchanged; a consumer mounting this under a pack must set it to 0, and 14 of the 67 snow-covered steps of the cold set have etp < 0, so it is not a hypothetical.

Observed

NameUnitsDefaultDescription
zemSoil layer interface DEPTHS below the surface, positive downward, ze[1] = 0; the negative of WRF’s ZSOIL prefixed by zero.
zcmSoil layer centre depths, from the grid’s column_cell_center template.
dzmSoil layer thicknesses, from the grid’s column_cell_thickness template. Equals WRF’s SLDPTH.
sh2om^3/m^3UNFROZEN volumetric soil water of each layer, WRF’s SH2O. This is the transported quantity: Richards’ equation moves liquid water only, and soil ice is an obstruction, not a participant.
smcm^3/m^3TOTAL volumetric soil moisture of each layer, WRF’s SMC, read only to form the ice content smc - sh2o.
etW/m^2Transpiration withdrawn from each soil layer, in energy units as the evapotranspiration component reports it. Converted internally to a water velocity.
prcp1m/sPrecipitation as a water velocity, prcp/rhowater.
edir1m/sDirect soil evaporation as a water velocity, edir/(lvh2o rhowater).
ec1m/sCanopy evaporation as a water velocity.
et1m/sPer-layer transpiration withdrawal as a water velocity.
rhsctm/sNet input to the canopy reservoir, shdfac prcp1 - ec1: interception minus evaporation, and negative when a dry-ish canopy is evaporating.
canopy_full1Indicator, 1 where the canopy reservoir is at or above capacity. WRF’s drip test cmc + dt rhsct > cmcmax becomes exactly this switch as dt -> 0.
dripm/sCanopy drip: zero while the canopy has room, and the whole positive net input once it is full. The dt -> 0 limit of WRF’s DRIP/dt.
dcmc_dtm/sTendency of the canopy water content, rhsct - drip. Zero for a full canopy that is gaining, which is what makes the reservoir hold at cmcmax.
pcpdrpm/sTotal water arriving at the soil surface, (1 - shdfac) prcp1 + drip.
sicem^3/m^3Volumetric soil ice of each layer, smc - sh2o. Zero everywhere in the present reference set.
sicemaxm^3/m^3Largest soil ice content in the column, which throttles the diffusivity in WDFCND. Untested: zero throughout the reference set.
factr21Degree of saturation in liquid water, sh2o/smcmax, the argument of both Clapp-Hornberger closures.
factr11min(0.05/smcmax, factr2 of the layer), the reduced saturation used for the ice-blocked part of the diffusivity. Carried per layer below.
vkwgt1Ice weighting of the diffusivity, 1/(1 + (500 sicemax)^3): one for ice-free soil, falling towards zero as ice fills the pores.
wdfm^2/sSoil water diffusivity of each layer, dwsat factr2^(bexp+2), blended by vkwgt towards the ice-throttled form where soil ice is present.
wcndm/sHydraulic conductivity of each layer, dksat factr2^(2 bexp + 3). This is the gravity drainage velocity.
smcavm^3/m^3Available pore space between wilting point and saturation, smcmax - smcwlt.
dmax_densitym^3/m^3Per-layer storage deficit per unit depth, smcav - (sh2o + sice - smcwlt), i.e. how much more water this layer could hold. Integrating it over the column gives the depth of water the profile can still absorb.
ddmColumn-integrated storage deficit, the depth of water the profile can still absorb.
dice_densitym^3/m^3Soil ice per unit depth, the integrand of dice.
dicemColumn-integrated soil ice depth, which gates the frozen-ground infiltration reduction at 1e-2 m.
r_infm/sStorage-limited infiltration rate, dd kdt/86400. The 86400 is WRF converting its kdt, which is defined per day, into per second; it appears because the Fortran forms dt1 = dt/86400.
infmax_storagem/sStorage-limited infiltration capacity in the dt -> 0 limit, pcpdrp r_inf/(pcpdrp + r_inf): a harmonic-style combination that tends to the supply for a dry profile and to r_inf for a wet one.
acrt1Frozen-ground argument, 3 frzx/dice.
fcr1Frozen-ground infiltration reduction, 1 - exp(-acrt)(1 + acrt + acrt^2/2), and exactly 1 where the column ice depth is at or below 1e-2 m. The polynomial is WRF’s CVFRZ = 3 factorial series written out. UNTESTED: no state in the reference set has soil ice.
infmaxm/sInfiltration capacity: the storage limit reduced by fcr, floored at the surface layer’s own conductivity (a soil can always take water at least as fast as it conducts it) and capped at the supply.
pddumm/sWater that actually enters the soil, min(pcpdrp, infmax). The surface boundary flux of the PDE.
runoff1m/sSurface runoff, the supply the soil could not take, max(pcpdrp - infmax, 0).
runoff2m/sSubsurface runoff (baseflow), slope wcnd[NLEV]: the gravity drainage out of the bottom of the column, reduced by the slope factor. Note that WRF ADDS runoff3/dt to this before reporting it, so a consumer comparing against WRF’s RUNOFF2 must add the supersaturation term this component does not produce.
K_wm^2/sSoil water diffusivity at each interface, taken from the layer ABOVE it, K_w[kn] = wdf[kn-1] (FORTRAN_BUGS N90). The entries at the two domain faces are never read: the lowering takes the prescribed F_w there instead.
F_wm/sPrescribed water flux at each interface, positive DOWNWARD: the infiltration pddum at kn = 1, the UPWIND gravity drainage wcnd[kn-1] in the interior, and the slope-reduced free drainage at kn = NLEV+1. Putting gravity here rather than in the diffusive term is what lets the existing face-flux rule serve Richards’ equation unchanged.
sink_density1/sWater withdrawn from each layer per unit depth: transpiration everywhere, plus bare-soil evaporation in the top layer only.
dsh2o_dt1/sInstantaneous tendency of the unfrozen soil water, the flux divergence minus the withdrawals. The dt -> 0 limit of WRF’s (SH2O_out - SH2O_in)/dt, whose finite-dt value differs because SSTEP solves the same operator implicitly.
dsmc_dt1/sInstantaneous tendency of the TOTAL soil moisture. Equal to dsh2o_dt because nothing here freezes or melts: the phase-change sink lives in SNKSRC, which is not transcribed. A document that later adds the freezing branch must make these two differ.
supersaturation_active1Indicator, 1 where any layer is at or above smcmax and still gaining water, which is the condition under which SSTEP’s untranscribed supersaturation redistribution would fire and produce runoff3. Zero throughout the reference set. A consumer that sees this go to one is outside what this component reproduces.
dewm/sDew deposition rate, max(-etp, 0)/(lvh2o rhowater). WRF’s NOPAC sets DEW = -ETP1 and then PRCP1 = PRCP1 + DEW on the etp <= 0 branch, so dew is routed exactly like rainfall: split between canopy and throughfall by shdfac. Missing this term leaves a small positive canopy tendency unexplained on every dew step.
prcp1_effm/sTotal liquid water supplied to the surface, prcp1 + dew_to_soildew + melt. This, not prcp1, is what the canopy and the soil see. The three terms are the three ways WRF gets water to the surface and they enter at the same point: rain (or, under a pack, rain that falls as rain), frost that NOPAC routes into the soil, and snowmelt that SNOPAC routes there. Verified against the reference: prcpf0.001 + ex reproduces SNOPAC’s own PRCP1 bitwise at all 67 snow-covered steps.
z_sfcmLower boundary of the column, ze[1]: the free name the EarthSciDiscretizations column_nonuniform_1d integral rules match as the lower bound of a bottom-anchored integral (ESD #39).
z_topmUpper boundary of the column, ze[NLEV+1]: the free name the EarthSciDiscretizations column_nonuniform_1d integral rules match as the upper bound of a top-anchored integral (ESD #39).

Observed expressions

$$ ze = \mathrm{input_ze}\left(\right) $$

Soil layer interface DEPTHS below the surface, positive downward, ze[1] = 0; the negative of WRF’s ZSOIL prefixed by zero.

$$ zc = \mathrm{column_cell_center}\left(ze=ze\right) $$

Soil layer centre depths, from the grid’s column_cell_center template.

$$ dz = \mathrm{column_cell_thickness}\left(ze=ze\right) $$

Soil layer thicknesses, from the grid’s column_cell_thickness template. Equals WRF’s SLDPTH.

$$ sh2o = \mathrm{input_sh2o}\left(\right) $$

UNFROZEN volumetric soil water of each layer, WRF’s SH2O. This is the transported quantity: Richards’ equation moves liquid water only, and soil ice is an obstruction, not a participant.

$$ smc = \mathrm{input_smc}\left(\right) $$

TOTAL volumetric soil moisture of each layer, WRF’s SMC, read only to form the ice content smc - sh2o.

$$ et = \mathrm{input_et}\left(\right) $$

Transpiration withdrawn from each soil layer, in energy units as the evapotranspiration component reports it. Converted internally to a water velocity.

$$ prcp1 = \frac{prcp}{wrf.rhowater} $$

Precipitation as a water velocity, prcp/rhowater.

$$ edir1 = \frac{edir}{par.lvh2o \cdot wrf.rhowater} $$

Direct soil evaporation as a water velocity, edir/(lvh2o rhowater).

$$ ec1 = \frac{ec}{par.lvh2o \cdot wrf.rhowater} $$

Canopy evaporation as a water velocity.

$$ et1 = \frac{et}{par.lvh2o \cdot wrf.rhowater} $$

Per-layer transpiration withdrawal as a water velocity.

$$ rhsct = shdfac \cdot prcp1_{eff} - ec1 $$

Net input to the canopy reservoir, shdfac prcp1 - ec1: interception minus evaporation, and negative when a dry-ish canopy is evaporating.

$$ canopy_{full} = \mathrm{ifelse}\left(\mathrm{>=}\left(cmc, par.cmcmax\right), 1.0, 0.0\right) $$

Indicator, 1 where the canopy reservoir is at or above capacity. WRF’s drip test cmc + dt rhsct > cmcmax becomes exactly this switch as dt -> 0.

$$ drip = \mathrm{ifelse}\left(canopy_{full} = 1.0, \mathrm{max}\left(rhsct, 0.0\right), 0.0\right) $$

Canopy drip: zero while the canopy has room, and the whole positive net input once it is full. The dt -> 0 limit of WRF’s DRIP/dt.

$$ dcmc_{dt} = rhsct - drip $$

Tendency of the canopy water content, rhsct - drip. Zero for a full canopy that is gaining, which is what makes the reservoir hold at cmcmax.

$$ pcpdrp = \left(1.0 - shdfac\right) \cdot prcp1_{eff} + drip $$

Total water arriving at the soil surface, (1 - shdfac) prcp1 + drip.

$$ sice = smc - sh2o $$

Volumetric soil ice of each layer, smc - sh2o. Zero everywhere in the present reference set.

$$ sicemax = \mathrm{faq}\left(sice\right) $$

Largest soil ice content in the column, which throttles the diffusivity in WDFCND. Untested: zero throughout the reference set.

$$ factr2 = \frac{sh2o}{smcmax} $$

Degree of saturation in liquid water, sh2o/smcmax, the argument of both Clapp-Hornberger closures.

$$ factr1 = \frac{0.05}{smcmax} $$

min(0.05/smcmax, factr2 of the layer), the reduced saturation used for the ice-blocked part of the diffusivity. Carried per layer below.

$$ vkwgt = \frac{1.0}{1.0 + \left(500.0 \cdot sicemax\right)^{3.0}} $$

Ice weighting of the diffusivity, 1/(1 + (500 sicemax)^3): one for ice-free soil, falling towards zero as ice fills the pores.

$$ wdf = \mathrm{ifelse}\left(\mathrm{>}\left(sicemax, 0.0\right), vkwgt \cdot dwsat \cdot factr2^{bexp + 2.0} + \left(1.0 - vkwgt\right) \cdot dwsat \cdot \mathrm{min}\left(factr1, factr2\right)^{bexp + 2.0}, dwsat \cdot factr2^{bexp + 2.0}\right) $$

Soil water diffusivity of each layer, dwsat factr2^(bexp+2), blended by vkwgt towards the ice-throttled form where soil ice is present.

$$ wcnd = dksat \cdot factr2^{2.0 \cdot bexp + 3.0} $$

Hydraulic conductivity of each layer, dksat factr2^(2 bexp + 3). This is the gravity drainage velocity.

$$ smcav = smcmax - smcwlt $$

Available pore space between wilting point and saturation, smcmax - smcwlt.

$$ dmax_{density} = smcav - \left(sh2o + sice - smcwlt\right) $$

Per-layer storage deficit per unit depth, smcav - (sh2o + sice - smcwlt), i.e. how much more water this layer could hold. Integrating it over the column gives the depth of water the profile can still absorb.

$$ dd = \mathrm{integral}\left(dmax_{density}\right) $$

Column-integrated storage deficit, the depth of water the profile can still absorb.

$$ dice_{density} = sice $$

Soil ice per unit depth, the integrand of dice.

$$ dice = \mathrm{integral}\left(dice_{density}\right) $$

Column-integrated soil ice depth, which gates the frozen-ground infiltration reduction at 1e-2 m.

$$ r_{inf} = \frac{dd \cdot kdt}{86400.0} $$

Storage-limited infiltration rate, dd kdt/86400. The 86400 is WRF converting its kdt, which is defined per day, into per second; it appears because the Fortran forms dt1 = dt/86400.

$$ infmax_{storage} = \mathrm{ifelse}\left(\mathrm{>}\left(pcpdrp + r_{inf}, 0.0\right), \frac{pcpdrp \cdot r_{inf}}{pcpdrp + r_{inf}}, 0.0\right) $$

Storage-limited infiltration capacity in the dt -> 0 limit, pcpdrp r_inf/(pcpdrp + r_inf): a harmonic-style combination that tends to the supply for a dry profile and to r_inf for a wet one.

$$ acrt = \frac{3.0 \cdot frzx}{\mathrm{max}\left(dice, 0.01\right)} $$

Frozen-ground argument, 3 frzx/dice.

$$ fcr = \mathrm{ifelse}\left(\mathrm{>}\left(dice, 0.01\right), 1.0 - e^{0.0 - acrt} \cdot \left(1.0 + acrt + \frac{acrt^{2.0}}{2.0}\right), 1.0\right) $$

Frozen-ground infiltration reduction, 1 - exp(-acrt)(1 + acrt + acrt^2/2), and exactly 1 where the column ice depth is at or below 1e-2 m. The polynomial is WRF’s CVFRZ = 3 factorial series written out. UNTESTED: no state in the reference set has soil ice.

$$ infmax = \mathrm{min}\left(\mathrm{max}\left(infmax_{storage} \cdot fcr, \mathrm{index}\left(wcnd, 1\right)\right), pcpdrp\right) $$

Infiltration capacity: the storage limit reduced by fcr, floored at the surface layer’s own conductivity (a soil can always take water at least as fast as it conducts it) and capped at the supply.

$$ pddum = \mathrm{min}\left(pcpdrp, infmax\right) $$

Water that actually enters the soil, min(pcpdrp, infmax). The surface boundary flux of the PDE.

$$ runoff1 = \mathrm{max}\left(pcpdrp - infmax, 0.0\right) $$

Surface runoff, the supply the soil could not take, max(pcpdrp - infmax, 0).

$$ runoff2 = slope \cdot \mathrm{index}\left(wcnd, NLEV\right) $$

Subsurface runoff (baseflow), slope wcnd[NLEV]: the gravity drainage out of the bottom of the column, reduced by the slope factor. Note that WRF ADDS runoff3/dt to this before reporting it, so a consumer comparing against WRF’s RUNOFF2 must add the supersaturation term this component does not produce.

$$ K_{w} = \mathrm{faq}\left(wdf\right) $$

Soil water diffusivity at each interface, taken from the layer ABOVE it, K_w[kn] = wdf[kn-1] (FORTRAN_BUGS N90). The entries at the two domain faces are never read: the lowering takes the prescribed F_w there instead.

$$ F_{w} = \mathrm{faq}\left(wcnd\right) $$

Prescribed water flux at each interface, positive DOWNWARD: the infiltration pddum at kn = 1, the UPWIND gravity drainage wcnd[kn-1] in the interior, and the slope-reduced free drainage at kn = NLEV+1. Putting gravity here rather than in the diffusive term is what lets the existing face-flux rule serve Richards’ equation unchanged.

$$ sink_{density} = \mathrm{faq}\left(et1, dz\right) $$

Water withdrawn from each layer per unit depth: transpiration everywhere, plus bare-soil evaporation in the top layer only.

$$ dsh2o_{dt} = \mathrm{D}\left(K_{w} \cdot \mathrm{D}\left(sh2o\right) - F_{w}\right) - sink_{density} $$

Instantaneous tendency of the unfrozen soil water, the flux divergence minus the withdrawals. The dt -> 0 limit of WRF’s (SH2O_out - SH2O_in)/dt, whose finite-dt value differs because SSTEP solves the same operator implicitly.

$$ dsmc_{dt} = dsh2o_{dt} $$

Instantaneous tendency of the TOTAL soil moisture. Equal to dsh2o_dt because nothing here freezes or melts: the phase-change sink lives in SNKSRC, which is not transcribed. A document that later adds the freezing branch must make these two differ.

$$ supersaturation_{active} = \mathrm{faq}\left(smc, dsh2o_{dt}\right) $$

Indicator, 1 where any layer is at or above smcmax and still gaining water, which is the condition under which SSTEP’s untranscribed supersaturation redistribution would fire and produce runoff3. Zero throughout the reference set. A consumer that sees this go to one is outside what this component reproduces.

$$ dew = \frac{\mathrm{max}\left(0.0 - etp, 0.0\right)}{par.lvh2o \cdot wrf.rhowater} $$

Dew deposition rate, max(-etp, 0)/(lvh2o rhowater). WRF’s NOPAC sets DEW = -ETP1 and then PRCP1 = PRCP1 + DEW on the etp <= 0 branch, so dew is routed exactly like rainfall: split between canopy and throughfall by shdfac. Missing this term leaves a small positive canopy tendency unexplained on every dew step.

$$ prcp1_{eff} = prcp1 + dew_{to\,soil} \cdot dew + melt $$

Total liquid water supplied to the surface, prcp1 + dew_to_soildew + melt. This, not prcp1, is what the canopy and the soil see. The three terms are the three ways WRF gets water to the surface and they enter at the same point: rain (or, under a pack, rain that falls as rain), frost that NOPAC routes into the soil, and snowmelt that SNOPAC routes there. Verified against the reference: prcpf0.001 + ex reproduces SNOPAC’s own PRCP1 bitwise at all 67 snow-covered steps.

$$ z_{sfc} = \mathrm{index}\left(ze, 1\right) $$

Lower boundary of the column, ze[1]: the free name the EarthSciDiscretizations column_nonuniform_1d integral rules match as the lower bound of a bottom-anchored integral (ESD #39).

$$ z_{top} = \mathrm{index}\left(ze, NLEV + 1\right) $$

Upper boundary of the column, ze[NLEV+1]: the free name the EarthSciDiscretizations column_nonuniform_1d integral rules match as the upper bound of a top-anchored integral (ESD #39).

Raw .esm

Raw .esm JSON (this component)
{
  "expression_template_imports": [
    {
      "ref": "${ESD_ROOT}/grids/column_nonuniform_1d/grid.esm",
      "bindings": {
        "NLEV": "NLEV"
      }
    },
    {
      "ref": "${ESD_ROOT}/grids/column_nonuniform_1d/rules/varcoeff_face_flux_laplacian_lev.esm",
      "bindings": {
        "NLEV": "NLEV"
      }
    },
    {
      "ref": "${ESD_ROOT}/grids/column_nonuniform_1d/rules/integral_lev_whole.esm",
      "bindings": {
        "NLEV": "NLEV"
      }
    }
  ],
  "subsystems": {
    "wrf": {
      "ref": "../../../lib/wrf_constants.esm"
    },
    "par": {
      "ref": "./noah_parameters.esm"
    }
  },
  "variables": {
    "ze": {
      "type": "unknown",
      "units": "m",
      "shape": [
        "lev_nodes"
      ],
      "description": "Soil layer interface DEPTHS below the surface, positive downward, ze[1] = 0; the negative of WRF's ZSOIL prefixed by zero."
    },
    "zc": {
      "type": "unknown",
      "units": "m",
      "shape": [
        "lev"
      ],
      "description": "Soil layer centre depths, from the grid's column_cell_center template."
    },
    "dz": {
      "type": "unknown",
      "units": "m",
      "shape": [
        "lev"
      ],
      "description": "Soil layer thicknesses, from the grid's column_cell_thickness template. Equals WRF's SLDPTH."
    },
    "sh2o": {
      "type": "unknown",
      "units": "m^3/m^3",
      "shape": [
        "lev"
      ],
      "description": "UNFROZEN volumetric soil water of each layer, WRF's SH2O. This is the transported quantity: Richards' equation moves liquid water only, and soil ice is an obstruction, not a participant."
    },
    "smc": {
      "type": "unknown",
      "units": "m^3/m^3",
      "shape": [
        "lev"
      ],
      "description": "TOTAL volumetric soil moisture of each layer, WRF's SMC, read only to form the ice content smc - sh2o."
    },
    "sh2o_in": {
      "type": "parameter",
      "units": "m^3/m^3",
      "shape": [
        "lev"
      ],
      "default": 0.0,
      "description": "Coupling target (assembly input path): a shaped parameter that receives the unfrozen soil water from an assembly's ODE state through a document-level variable_map (param_to_var), when that document's mount-edge library lowers `input_sh2o` to it (esm-spec 9.7.10, 10.4; couplings/noah_column.esm). Unused by this component's own inline tests, which lower `input_sh2o` to dumped constants."
    },
    "smc_in": {
      "type": "parameter",
      "units": "m^3/m^3",
      "shape": [
        "lev"
      ],
      "default": 0.0,
      "description": "Coupling target for the total soil moisture, as sh2o_in (lowered from `input_smc` by an assembly's mount-edge library)."
    },
    "et_in": {
      "type": "parameter",
      "units": "W/m^2",
      "shape": [
        "lev"
      ],
      "default": 0.0,
      "description": "Coupling target for the per-layer transpiration, which an assembly hands over from the evapotranspiration component's et through a document-level variable_map (param_to_var) when its mount-edge library lowers `input_et` to it."
    },
    "et": {
      "type": "unknown",
      "units": "W/m^2",
      "shape": [
        "lev"
      ],
      "description": "Transpiration withdrawn from each soil layer, in energy units as the evapotranspiration component reports it. Converted internally to a water velocity."
    },
    "edir": {
      "type": "parameter",
      "units": "W/m^2",
      "default": 30.0,
      "description": "Direct soil evaporation in energy units, which withdraws from the TOP LAYER ONLY."
    },
    "ec": {
      "type": "parameter",
      "units": "W/m^2",
      "default": 50.0,
      "description": "Canopy-water evaporation in energy units, which drains the canopy reservoir and never touches the soil."
    },
    "prcp": {
      "type": "parameter",
      "units": "kg/(m^2*s)",
      "default": 0.0,
      "description": "Precipitation rate reaching the surface, WRF's PRCP."
    },
    "shdfac": {
      "type": "parameter",
      "units": "1",
      "default": 0.5,
      "description": "Green vegetation fraction, which splits the precipitation between canopy interception (shdfac) and direct throughfall (1 - shdfac)."
    },
    "cmc": {
      "type": "parameter",
      "units": "m",
      "default": 0.0005,
      "description": "Canopy-intercepted water content, WRF's CMC."
    },
    "smcmax": {
      "type": "parameter",
      "units": "m^3/m^3",
      "default": 0.476,
      "description": "Soil porosity, SOILPARM.TBL MAXSMC."
    },
    "smcwlt": {
      "type": "parameter",
      "units": "m^3/m^3",
      "default": 0.084,
      "description": "Soil wilting point, SOILPARM.TBL WLTSMC, which sets the bottom of the storage deficit the infiltration limiter measures."
    },
    "bexp": {
      "type": "parameter",
      "units": "1",
      "default": 5.33,
      "description": "Clapp-Hornberger b, SOILPARM.TBL BB."
    },
    "dksat": {
      "type": "parameter",
      "units": "m/s",
      "default": 2.81e-06,
      "description": "Saturated hydraulic conductivity, SOILPARM.TBL SATDK."
    },
    "dwsat": {
      "type": "parameter",
      "units": "m^2/s",
      "default": 2.39e-05,
      "description": "Saturated soil water diffusivity, SOILPARM.TBL SATDW."
    },
    "kdt": {
      "type": "parameter",
      "units": "1",
      "default": 4.215,
      "description": "Infiltration parameter, refkdt dksat/refdk from REDPRM."
    },
    "frzx": {
      "type": "parameter",
      "units": "1",
      "default": 0.1748,
      "description": "Per-soil-type frozen-ground parameter, frzk frzfact from REDPRM. Read only by the fcr reduction, which no state in the reference set reaches."
    },
    "slope": {
      "type": "parameter",
      "units": "1",
      "default": 0.1,
      "description": "Bottom drainage-slope factor, GENPARM.TBL SLOPE_DATA: the fraction of the deepest layer's gravity drainage that actually leaves the column as subsurface runoff."
    },
    "prcp1": {
      "type": "unknown",
      "units": "m/s",
      "description": "Precipitation as a water velocity, prcp/rhowater."
    },
    "edir1": {
      "type": "unknown",
      "units": "m/s",
      "description": "Direct soil evaporation as a water velocity, edir/(lvh2o rhowater)."
    },
    "ec1": {
      "type": "unknown",
      "units": "m/s",
      "description": "Canopy evaporation as a water velocity."
    },
    "et1": {
      "type": "unknown",
      "units": "m/s",
      "shape": [
        "lev"
      ],
      "description": "Per-layer transpiration withdrawal as a water velocity."
    },
    "rhsct": {
      "type": "unknown",
      "units": "m/s",
      "description": "Net input to the canopy reservoir, shdfac prcp1 - ec1: interception minus evaporation, and negative when a dry-ish canopy is evaporating."
    },
    "canopy_full": {
      "type": "unknown",
      "units": "1",
      "description": "Indicator, 1 where the canopy reservoir is at or above capacity. WRF's drip test cmc + dt rhsct > cmcmax becomes exactly this switch as dt -> 0."
    },
    "drip": {
      "type": "unknown",
      "units": "m/s",
      "description": "Canopy drip: zero while the canopy has room, and the whole positive net input once it is full. The dt -> 0 limit of WRF's DRIP/dt."
    },
    "dcmc_dt": {
      "type": "unknown",
      "units": "m/s",
      "description": "Tendency of the canopy water content, rhsct - drip. Zero for a full canopy that is gaining, which is what makes the reservoir hold at cmcmax."
    },
    "pcpdrp": {
      "type": "unknown",
      "units": "m/s",
      "description": "Total water arriving at the soil surface, (1 - shdfac) prcp1 + drip."
    },
    "sice": {
      "type": "unknown",
      "units": "m^3/m^3",
      "shape": [
        "lev"
      ],
      "description": "Volumetric soil ice of each layer, smc - sh2o. Zero everywhere in the present reference set."
    },
    "sicemax": {
      "type": "unknown",
      "units": "m^3/m^3",
      "description": "Largest soil ice content in the column, which throttles the diffusivity in WDFCND. Untested: zero throughout the reference set."
    },
    "factr2": {
      "type": "unknown",
      "units": "1",
      "shape": [
        "lev"
      ],
      "description": "Degree of saturation in liquid water, sh2o/smcmax, the argument of both Clapp-Hornberger closures."
    },
    "factr1": {
      "type": "unknown",
      "units": "1",
      "description": "min(0.05/smcmax, factr2 of the layer), the reduced saturation used for the ice-blocked part of the diffusivity. Carried per layer below."
    },
    "vkwgt": {
      "type": "unknown",
      "units": "1",
      "description": "Ice weighting of the diffusivity, 1/(1 + (500 sicemax)^3): one for ice-free soil, falling towards zero as ice fills the pores."
    },
    "wdf": {
      "type": "unknown",
      "units": "m^2/s",
      "shape": [
        "lev"
      ],
      "description": "Soil water diffusivity of each layer, dwsat factr2^(bexp+2), blended by vkwgt towards the ice-throttled form where soil ice is present."
    },
    "wcnd": {
      "type": "unknown",
      "units": "m/s",
      "shape": [
        "lev"
      ],
      "description": "Hydraulic conductivity of each layer, dksat factr2^(2 bexp + 3). This is the gravity drainage velocity."
    },
    "smcav": {
      "type": "unknown",
      "units": "m^3/m^3",
      "description": "Available pore space between wilting point and saturation, smcmax - smcwlt."
    },
    "dmax_density": {
      "type": "unknown",
      "units": "m^3/m^3",
      "shape": [
        "lev"
      ],
      "description": "Per-layer storage deficit per unit depth, smcav - (sh2o + sice - smcwlt), i.e. how much more water this layer could hold. Integrating it over the column gives the depth of water the profile can still absorb."
    },
    "dd": {
      "type": "unknown",
      "units": "m",
      "description": "Column-integrated storage deficit, the depth of water the profile can still absorb."
    },
    "dice_density": {
      "type": "unknown",
      "units": "m^3/m^3",
      "shape": [
        "lev"
      ],
      "description": "Soil ice per unit depth, the integrand of dice."
    },
    "dice": {
      "type": "unknown",
      "units": "m",
      "description": "Column-integrated soil ice depth, which gates the frozen-ground infiltration reduction at 1e-2 m."
    },
    "r_inf": {
      "type": "unknown",
      "units": "m/s",
      "description": "Storage-limited infiltration rate, dd kdt/86400. The 86400 is WRF converting its kdt, which is defined per day, into per second; it appears because the Fortran forms dt1 = dt/86400."
    },
    "infmax_storage": {
      "type": "unknown",
      "units": "m/s",
      "description": "Storage-limited infiltration capacity in the dt -> 0 limit, pcpdrp r_inf/(pcpdrp + r_inf): a harmonic-style combination that tends to the supply for a dry profile and to r_inf for a wet one."
    },
    "acrt": {
      "type": "unknown",
      "units": "1",
      "description": "Frozen-ground argument, 3 frzx/dice."
    },
    "fcr": {
      "type": "unknown",
      "units": "1",
      "description": "Frozen-ground infiltration reduction, 1 - exp(-acrt)(1 + acrt + acrt^2/2), and exactly 1 where the column ice depth is at or below 1e-2 m. The polynomial is WRF's CVFRZ = 3 factorial series written out. UNTESTED: no state in the reference set has soil ice."
    },
    "infmax": {
      "type": "unknown",
      "units": "m/s",
      "description": "Infiltration capacity: the storage limit reduced by fcr, floored at the surface layer's own conductivity (a soil can always take water at least as fast as it conducts it) and capped at the supply."
    },
    "pddum": {
      "type": "unknown",
      "units": "m/s",
      "description": "Water that actually enters the soil, min(pcpdrp, infmax). The surface boundary flux of the PDE."
    },
    "runoff1": {
      "type": "unknown",
      "units": "m/s",
      "description": "Surface runoff, the supply the soil could not take, max(pcpdrp - infmax, 0)."
    },
    "runoff2": {
      "type": "unknown",
      "units": "m/s",
      "description": "Subsurface runoff (baseflow), slope wcnd[NLEV]: the gravity drainage out of the bottom of the column, reduced by the slope factor. Note that WRF ADDS runoff3/dt to this before reporting it, so a consumer comparing against WRF's RUNOFF2 must add the supersaturation term this component does not produce."
    },
    "K_w": {
      "type": "unknown",
      "units": "m^2/s",
      "shape": [
        "lev_nodes"
      ],
      "description": "Soil water diffusivity at each interface, taken from the layer ABOVE it, K_w[kn] = wdf[kn-1] (FORTRAN_BUGS N90). The entries at the two domain faces are never read: the lowering takes the prescribed F_w there instead."
    },
    "F_w": {
      "type": "unknown",
      "units": "m/s",
      "shape": [
        "lev_nodes"
      ],
      "description": "Prescribed water flux at each interface, positive DOWNWARD: the infiltration pddum at kn = 1, the UPWIND gravity drainage wcnd[kn-1] in the interior, and the slope-reduced free drainage at kn = NLEV+1. Putting gravity here rather than in the diffusive term is what lets the existing face-flux rule serve Richards' equation unchanged."
    },
    "sink_density": {
      "type": "unknown",
      "units": "1/s",
      "shape": [
        "lev"
      ],
      "description": "Water withdrawn from each layer per unit depth: transpiration everywhere, plus bare-soil evaporation in the top layer only."
    },
    "dsh2o_dt": {
      "type": "unknown",
      "units": "1/s",
      "shape": [
        "lev"
      ],
      "description": "Instantaneous tendency of the unfrozen soil water, the flux divergence minus the withdrawals. The dt -> 0 limit of WRF's (SH2O_out - SH2O_in)/dt, whose finite-dt value differs because SSTEP solves the same operator implicitly."
    },
    "dsmc_dt": {
      "type": "unknown",
      "units": "1/s",
      "shape": [
        "lev"
      ],
      "description": "Instantaneous tendency of the TOTAL soil moisture. Equal to dsh2o_dt because nothing here freezes or melts: the phase-change sink lives in SNKSRC, which is not transcribed. A document that later adds the freezing branch must make these two differ."
    },
    "supersaturation_active": {
      "type": "unknown",
      "units": "1",
      "description": "Indicator, 1 where any layer is at or above smcmax and still gaining water, which is the condition under which SSTEP's untranscribed supersaturation redistribution would fire and produce runoff3. Zero throughout the reference set. A consumer that sees this go to one is outside what this component reproduces."
    },
    "etp": {
      "type": "parameter",
      "units": "W/m^2",
      "default": 100.0,
      "description": "Potential evaporation in energy units from the Penman component. Read ONLY for its sign: where it is negative the surface is gaining dew, and NOPAC adds that gain to the precipitation reaching the canopy and the soil."
    },
    "dew": {
      "type": "unknown",
      "units": "m/s",
      "description": "Dew deposition rate, max(-etp, 0)/(lvh2o rhowater). WRF's NOPAC sets DEW = -ETP1 and then PRCP1 = PRCP1 + DEW on the etp <= 0 branch, so dew is routed exactly like rainfall: split between canopy and throughfall by shdfac. Missing this term leaves a small positive canopy tendency unexplained on every dew step."
    },
    "prcp1_eff": {
      "type": "unknown",
      "units": "m/s",
      "description": "Total liquid water supplied to the surface, prcp1 + dew_to_soil*dew + melt. This, not prcp1, is what the canopy and the soil see. The three terms are the three ways WRF gets water to the surface and they enter at the same point: rain (or, under a pack, rain that falls as rain), frost that NOPAC routes into the soil, and snowmelt that SNOPAC routes there. Verified against the reference: prcpf*0.001 + ex reproduces SNOPAC's own PRCP1 bitwise at all 67 snow-covered steps."
    },
    "melt": {
      "type": "parameter",
      "units": "m/s",
      "default": 0.0,
      "description": "Snowmelt water arriving at the surface, WRF's EX, which SNOPAC adds to PRCP1 before calling SMFLX. Zero on a snow-free column, where SFLX takes the NOPAC branch and there is no pack to melt. It joins the supply at exactly the same point as rain and therefore goes through the same canopy split; it is NOT a separate soil source. On a melting pack it dominates everything else: 8.0e-7 m/s at the strongest step of the cold-season set, against a bare-ground column's dew of order 1e-9."
    },
    "dew_to_soil": {
      "type": "parameter",
      "units": "1",
      "default": 1.0,
      "description": "Whether dew and frost are routed into the SOIL water, 1 on a snow-free column and 0 under a snow pack. This is a real structural difference between WRF's two branches and not a switch of convenience: NOPAC does PRCP1 = PRCP1 + DEW, so frost on bare ground infiltrates, while SNOPAC sets DEW = -ETP1 and never adds it to PRCP1 -- the frost is deposited on the PACK and reaches the soil only later, as melt. Defaulting to 1 keeps every snow-free test unchanged; a consumer mounting this under a pack must set it to 0, and 14 of the 67 snow-covered steps of the cold set have etp < 0, so it is not a hypothetical."
    },
    "z_sfc": {
      "type": "unknown",
      "units": "m",
      "description": "Lower boundary of the column, ze[1]: the free name the EarthSciDiscretizations column_nonuniform_1d integral rules match as the lower bound of a bottom-anchored integral (ESD #39)."
    },
    "z_top": {
      "type": "unknown",
      "units": "m",
      "description": "Upper boundary of the column, ze[NLEV+1]: the free name the EarthSciDiscretizations column_nonuniform_1d integral rules match as the upper bound of a top-anchored integral (ESD #39)."
    }
  },
  "equations": [
    {
      "lhs": "ze",
      "rhs": {
        "op": "input_ze",
        "args": []
      }
    },
    {
      "lhs": "zc",
      "rhs": {
        "op": "apply_expression_template",
        "args": [],
        "name": "column_cell_center",
        "bindings": {
          "ze": "ze"
        }
      }
    },
    {
      "lhs": "dz",
      "rhs": {
        "op": "apply_expression_template",
        "args": [],
        "name": "column_cell_thickness",
        "bindings": {
          "ze": "ze"
        }
      }
    },
    {
      "lhs": "sh2o",
      "rhs": {
        "op": "input_sh2o",
        "args": []
      }
    },
    {
      "lhs": "smc",
      "rhs": {
        "op": "input_smc",
        "args": []
      }
    },
    {
      "lhs": "et",
      "rhs": {
        "op": "input_et",
        "args": []
      }
    },
    {
      "lhs": "prcp1",
      "rhs": {
        "op": "/",
        "args": [
          "prcp",
          "wrf.rhowater"
        ]
      }
    },
    {
      "lhs": "edir1",
      "rhs": {
        "op": "/",
        "args": [
          "edir",
          {
            "op": "*",
            "args": [
              "par.lvh2o",
              "wrf.rhowater"
            ]
          }
        ]
      }
    },
    {
      "lhs": "ec1",
      "rhs": {
        "op": "/",
        "args": [
          "ec",
          {
            "op": "*",
            "args": [
              "par.lvh2o",
              "wrf.rhowater"
            ]
          }
        ]
      }
    },
    {
      "lhs": "et1",
      "rhs": {
        "op": "/",
        "args": [
          "et",
          {
            "op": "*",
            "args": [
              "par.lvh2o",
              "wrf.rhowater"
            ]
          }
        ]
      }
    },
    {
      "lhs": "dew",
      "rhs": {
        "op": "/",
        "args": [
          {
            "op": "max",
            "args": [
              {
                "op": "-",
                "args": [
                  0.0,
                  "etp"
                ]
              },
              0.0
            ]
          },
          {
            "op": "*",
            "args": [
              "par.lvh2o",
              "wrf.rhowater"
            ]
          }
        ]
      }
    },
    {
      "lhs": "prcp1_eff",
      "rhs": {
        "op": "+",
        "args": [
          {
            "op": "+",
            "args": [
              "prcp1",
              {
                "op": "*",
                "args": [
                  "dew_to_soil",
                  "dew"
                ]
              }
            ]
          },
          "melt"
        ]
      }
    },
    {
      "lhs": "rhsct",
      "rhs": {
        "op": "-",
        "args": [
          {
            "op": "*",
            "args": [
              "shdfac",
              "prcp1_eff"
            ]
          },
          "ec1"
        ]
      }
    },
    {
      "lhs": "canopy_full",
      "rhs": {
        "op": "ifelse",
        "args": [
          {
            "op": ">=",
            "args": [
              "cmc",
              "par.cmcmax"
            ]
          },
          1.0,
          0.0
        ]
      }
    },
    {
      "lhs": "drip",
      "rhs": {
        "op": "ifelse",
        "args": [
          {
            "op": "==",
            "args": [
              "canopy_full",
              1.0
            ]
          },
          {
            "op": "max",
            "args": [
              "rhsct",
              0.0
            ]
          },
          0.0
        ]
      }
    },
    {
      "lhs": "dcmc_dt",
      "rhs": {
        "op": "-",
        "args": [
          "rhsct",
          "drip"
        ]
      }
    },
    {
      "lhs": "pcpdrp",
      "rhs": {
        "op": "+",
        "args": [
          {
            "op": "*",
            "args": [
              {
                "op": "-",
                "args": [
                  1.0,
                  "shdfac"
                ]
              },
              "prcp1_eff"
            ]
          },
          "drip"
        ]
      }
    },
    {
      "lhs": "sice",
      "rhs": {
        "op": "-",
        "args": [
          "smc",
          "sh2o"
        ]
      }
    },
    {
      "lhs": "sicemax",
      "rhs": {
        "op": "faq",
        "output_idx": [],
        "args": [
          "sice"
        ],
        "ranges": {
          "k": {
            "from": "lev"
          }
        },
        "reduce": "max",
        "expr": {
          "op": "max",
          "args": [
            {
              "op": "index",
              "args": [
                "sice",
                "k"
              ]
            },
            0.0
          ]
        }
      }
    },
    {
      "lhs": "factr2",
      "rhs": {
        "op": "/",
        "args": [
          "sh2o",
          "smcmax"
        ]
      }
    },
    {
      "lhs": "factr1",
      "rhs": {
        "op": "/",
        "args": [
          0.05,
          "smcmax"
        ]
      }
    },
    {
      "lhs": "vkwgt",
      "rhs": {
        "op": "/",
        "args": [
          1.0,
          {
            "op": "+",
            "args": [
              1.0,
              {
                "op": "^",
                "args": [
                  {
                    "op": "*",
                    "args": [
                      500.0,
                      "sicemax"
                    ]
                  },
                  3.0
                ]
              }
            ]
          }
        ]
      }
    },
    {
      "lhs": "wdf",
      "rhs": {
        "op": "ifelse",
        "args": [
          {
            "op": ">",
            "args": [
              "sicemax",
              0.0
            ]
          },
          {
            "op": "+",
            "args": [
              {
                "op": "*",
                "args": [
                  "vkwgt",
                  {
                    "op": "*",
                    "args": [
                      "dwsat",
                      {
                        "op": "^",
                        "args": [
                          "factr2",
                          {
                            "op": "+",
                            "args": [
                              "bexp",
                              2.0
                            ]
                          }
                        ]
                      }
                    ]
                  }
                ]
              },
              {
                "op": "*",
                "args": [
                  {
                    "op": "-",
                    "args": [
                      1.0,
                      "vkwgt"
                    ]
                  },
                  {
                    "op": "*",
                    "args": [
                      "dwsat",
                      {
                        "op": "^",
                        "args": [
                          {
                            "op": "min",
                            "args": [
                              "factr1",
                              "factr2"
                            ]
                          },
                          {
                            "op": "+",
                            "args": [
                              "bexp",
                              2.0
                            ]
                          }
                        ]
                      }
                    ]
                  }
                ]
              }
            ]
          },
          {
            "op": "*",
            "args": [
              "dwsat",
              {
                "op": "^",
                "args": [
                  "factr2",
                  {
                    "op": "+",
                    "args": [
                      "bexp",
                      2.0
                    ]
                  }
                ]
              }
            ]
          }
        ]
      }
    },
    {
      "lhs": "wcnd",
      "rhs": {
        "op": "*",
        "args": [
          "dksat",
          {
            "op": "^",
            "args": [
              "factr2",
              {
                "op": "+",
                "args": [
                  {
                    "op": "*",
                    "args": [
                      2.0,
                      "bexp"
                    ]
                  },
                  3.0
                ]
              }
            ]
          }
        ]
      }
    },
    {
      "lhs": "smcav",
      "rhs": {
        "op": "-",
        "args": [
          "smcmax",
          "smcwlt"
        ]
      }
    },
    {
      "lhs": "dmax_density",
      "rhs": {
        "op": "-",
        "args": [
          "smcav",
          {
            "op": "-",
            "args": [
              {
                "op": "+",
                "args": [
                  "sh2o",
                  "sice"
                ]
              },
              "smcwlt"
            ]
          }
        ]
      }
    },
    {
      "lhs": "dd",
      "rhs": {
        "op": "integral",
        "args": [
          "dmax_density"
        ],
        "var": "lev",
        "lower": "z_sfc",
        "upper": "z_top"
      }
    },
    {
      "lhs": "dice_density",
      "rhs": "sice"
    },
    {
      "lhs": "dice",
      "rhs": {
        "op": "integral",
        "args": [
          "dice_density"
        ],
        "var": "lev",
        "lower": "z_sfc",
        "upper": "z_top"
      }
    },
    {
      "lhs": "r_inf",
      "rhs": {
        "op": "/",
        "args": [
          {
            "op": "*",
            "args": [
              "dd",
              "kdt"
            ]
          },
          86400.0
        ]
      }
    },
    {
      "lhs": "infmax_storage",
      "rhs": {
        "op": "ifelse",
        "args": [
          {
            "op": ">",
            "args": [
              {
                "op": "+",
                "args": [
                  "pcpdrp",
                  "r_inf"
                ]
              },
              0.0
            ]
          },
          {
            "op": "/",
            "args": [
              {
                "op": "*",
                "args": [
                  "pcpdrp",
                  "r_inf"
                ]
              },
              {
                "op": "+",
                "args": [
                  "pcpdrp",
                  "r_inf"
                ]
              }
            ]
          },
          0.0
        ]
      }
    },
    {
      "lhs": "acrt",
      "rhs": {
        "op": "/",
        "args": [
          {
            "op": "*",
            "args": [
              3.0,
              "frzx"
            ]
          },
          {
            "op": "max",
            "args": [
              "dice",
              0.01
            ]
          }
        ]
      }
    },
    {
      "lhs": "fcr",
      "rhs": {
        "op": "ifelse",
        "args": [
          {
            "op": ">",
            "args": [
              "dice",
              0.01
            ]
          },
          {
            "op": "-",
            "args": [
              1.0,
              {
                "op": "*",
                "args": [
                  {
                    "op": "exp",
                    "args": [
                      {
                        "op": "-",
                        "args": [
                          0.0,
                          "acrt"
                        ]
                      }
                    ]
                  },
                  {
                    "op": "+",
                    "args": [
                      {
                        "op": "+",
                        "args": [
                          1.0,
                          "acrt"
                        ]
                      },
                      {
                        "op": "/",
                        "args": [
                          {
                            "op": "^",
                            "args": [
                              "acrt",
                              2.0
                            ]
                          },
                          2.0
                        ]
                      }
                    ]
                  }
                ]
              }
            ]
          },
          1.0
        ]
      }
    },
    {
      "lhs": "infmax",
      "rhs": {
        "op": "min",
        "args": [
          {
            "op": "max",
            "args": [
              {
                "op": "*",
                "args": [
                  "infmax_storage",
                  "fcr"
                ]
              },
              {
                "op": "index",
                "args": [
                  "wcnd",
                  1
                ]
              }
            ]
          },
          "pcpdrp"
        ]
      }
    },
    {
      "lhs": "pddum",
      "rhs": {
        "op": "min",
        "args": [
          "pcpdrp",
          "infmax"
        ]
      }
    },
    {
      "lhs": "runoff1",
      "rhs": {
        "op": "max",
        "args": [
          {
            "op": "-",
            "args": [
              "pcpdrp",
              "infmax"
            ]
          },
          0.0
        ]
      }
    },
    {
      "lhs": "runoff2",
      "rhs": {
        "op": "*",
        "args": [
          "slope",
          {
            "op": "index",
            "args": [
              "wcnd",
              "NLEV"
            ]
          }
        ]
      }
    },
    {
      "lhs": "K_w",
      "rhs": {
        "op": "faq",
        "output_idx": [
          "kn"
        ],
        "args": [
          "wdf"
        ],
        "ranges": {
          "kn": {
            "from": "lev_nodes"
          }
        },
        "expr": {
          "op": "ifelse",
          "args": [
            {
              "op": "==",
              "args": [
                "kn",
                1
              ]
            },
            {
              "op": "index",
              "args": [
                "wdf",
                1
              ]
            },
            {
              "op": "index",
              "args": [
                "wdf",
                {
                  "op": "-",
                  "args": [
                    "kn",
                    1
                  ]
                }
              ]
            }
          ]
        }
      }
    },
    {
      "lhs": "F_w",
      "rhs": {
        "op": "faq",
        "output_idx": [
          "kn"
        ],
        "args": [
          "wcnd"
        ],
        "ranges": {
          "kn": {
            "from": "lev_nodes"
          }
        },
        "expr": {
          "op": "ifelse",
          "args": [
            {
              "op": "==",
              "args": [
                "kn",
                1
              ]
            },
            "pddum",
            {
              "op": "ifelse",
              "args": [
                {
                  "op": "==",
                  "args": [
                    "kn",
                    {
                      "op": "+",
                      "args": [
                        "NLEV",
                        1
                      ]
                    }
                  ]
                },
                {
                  "op": "*",
                  "args": [
                    "slope",
                    {
                      "op": "index",
                      "args": [
                        "wcnd",
                        "NLEV"
                      ]
                    }
                  ]
                },
                {
                  "op": "index",
                  "args": [
                    "wcnd",
                    {
                      "op": "-",
                      "args": [
                        "kn",
                        1
                      ]
                    }
                  ]
                }
              ]
            }
          ]
        }
      }
    },
    {
      "lhs": "sink_density",
      "rhs": {
        "op": "faq",
        "output_idx": [
          "k"
        ],
        "args": [
          "et1",
          "dz"
        ],
        "ranges": {
          "k": {
            "from": "lev"
          }
        },
        "expr": {
          "op": "/",
          "args": [
            {
              "op": "ifelse",
              "args": [
                {
                  "op": "==",
                  "args": [
                    "k",
                    1
                  ]
                },
                {
                  "op": "+",
                  "args": [
                    {
                      "op": "index",
                      "args": [
                        "et1",
                        1
                      ]
                    },
                    "edir1"
                  ]
                },
                {
                  "op": "index",
                  "args": [
                    "et1",
                    "k"
                  ]
                }
              ]
            },
            {
              "op": "index",
              "args": [
                "dz",
                "k"
              ]
            }
          ]
        }
      }
    },
    {
      "lhs": "dsh2o_dt",
      "rhs": {
        "op": "-",
        "args": [
          {
            "op": "D",
            "args": [
              {
                "op": "-",
                "args": [
                  {
                    "op": "*",
                    "args": [
                      "K_w",
                      {
                        "op": "D",
                        "args": [
                          "sh2o"
                        ],
                        "wrt": "lev"
                      }
                    ]
                  },
                  "F_w"
                ]
              }
            ],
            "wrt": "lev"
          },
          "sink_density"
        ]
      }
    },
    {
      "lhs": "dsmc_dt",
      "rhs": "dsh2o_dt"
    },
    {
      "lhs": "supersaturation_active",
      "rhs": {
        "op": "faq",
        "output_idx": [],
        "args": [
          "smc",
          "dsh2o_dt"
        ],
        "ranges": {
          "k": {
            "from": "lev"
          }
        },
        "reduce": "max",
        "expr": {
          "op": "ifelse",
          "args": [
            {
              "op": "and",
              "args": [
                {
                  "op": ">=",
                  "args": [
                    {
                      "op": "index",
                      "args": [
                        "smc",
                        "k"
                      ]
                    },
                    "smcmax"
                  ]
                },
                {
                  "op": ">",
                  "args": [
                    {
                      "op": "index",
                      "args": [
                        "dsh2o_dt",
                        "k"
                      ]
                    },
                    0.0
                  ]
                }
              ]
            },
            1.0,
            0.0
          ]
        }
      }
    },
    {
      "lhs": "z_sfc",
      "rhs": {
        "op": "index",
        "args": [
          "ze",
          1
        ]
      }
    },
    {
      "lhs": "z_top",
      "rhs": {
        "op": "index",
        "args": [
          "ze",
          {
            "op": "+",
            "args": [
              "NLEV",
              1
            ]
          }
        ]
      }
    }
  ],
  "tests": [
    {
      "id": "chem170_scm_step1_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at pre-dawn, no rain: drainage and root uptake only. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 0.1790657287391653,
        "ec": 0.0,
        "etp": 1.1795247991217812,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step1_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.129462179181446e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step1_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at pre-dawn, no rain: drainage and root uptake only. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 0.1790657287391653,
        "ec": 0.0,
        "etp": 1.1795247991217812,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step1_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -5.224599641806549e-08,
                    -2.3542279237176444e-10,
                    -2.354172412566413e-10,
                    4.61651827876608e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -5.224599641806549e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": -2.3542279237176444e-10,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": -2.354172412566413e-10,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 4.61651827876608e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -5.224599641806549e-08,
                    -2.3542279237176444e-10,
                    -2.354172412566413e-10,
                    4.61651827876608e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step60_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at just after sunrise, no rain. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 1.049336164188505,
        "ec": 3.037743904512558,
        "etp": 6.085228146909409,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0004984006518498063,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step60_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.133159532077107e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step60_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at just after sunrise, no rain. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 1.049336164188505,
        "ec": 3.037743904512558,
        "etp": 6.085228146909409,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0004984006518498063,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step60_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -1.2146117119894717e-09,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -8.56628618040034e-07,
                    2.606742988131572e-07,
                    3.1994906724008843e-09,
                    4.6060932845648495e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -8.56628618040034e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": 2.606742988131572e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": 3.1994906724008843e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 4.6060932845648495e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -8.56628618040034e-07,
                    2.606742988131572e-07,
                    3.1994906724008843e-09,
                    4.6060932845648495e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step230_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at strong mid-morning evaporative demand, no rain. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 64.62431406974817,
        "ec": 120.76098492315288,
        "etp": 398.32075381039306,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.00018383044516667724,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step230_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.14382673795685e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step230_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at strong mid-morning evaporative demand, no rain. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 64.62431406974817,
        "ec": 120.76098492315288,
        "etp": 398.32075381039306,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.00018383044516667724,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step230_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -4.8285079938276826e-08,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -6.011391329163729e-07,
                    3.963025463349368e-08,
                    -1.9675938656149583e-08,
                    4.555511523562927e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -6.011391329163729e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": 3.963025463349368e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": -1.9675938656149583e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 4.555511523562927e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -6.011391329163729e-07,
                    3.963025463349368e-08,
                    -1.9675938656149583e-08,
                    4.555511523562927e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step364_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at local noon, light rain fully absorbed (the supply cap binds). Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 53.54266767794682,
        "ec": 3.1599182285098832,
        "etp": 342.72821971617634,
        "prcp": 2.5604208531149197e-06,
        "shdfac": 0.5,
        "cmc": 1.7001293883822655e-07,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step364_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.151969915360055e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step364_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at local noon, light rain fully absorbed (the supply cap binds). SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 53.54266767794682,
        "ec": 3.1599182285098832,
        "etp": 342.72821971617634,
        "prcp": 2.5604208531149197e-06,
        "shdfac": 0.5,
        "cmc": 1.7001293883822655e-07,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step364_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": 1.6748519915683818e-11,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -3.68813157791692e-07,
                    -6.347780634463618e-08,
                    -5.086883136939946e-08,
                    4.040690004813996e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -3.68813157791692e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": -6.347780634463618e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": -5.086883136939946e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 4.040690004813996e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -3.68813157791692e-07,
                    -6.347780634463618e-08,
                    -5.086883136939946e-08,
                    4.040690004813996e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step601_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at afternoon, the heaviest rain of the set, canopy EXACTLY FULL so drip carries the whole canopy input and the storage limit binds. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 30.678776706834583,
        "ec": 87.72652299304896,
        "etp": 175.45304598609786,
        "prcp": 0.0006315885111689568,
        "shdfac": 0.5,
        "cmc": 0.0005000000237487257,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step601_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 2.011113467495341e-08,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.164109025659591e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step601_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at afternoon, the heaviest rain of the set, canopy EXACTLY FULL so drip carries the whole canopy input and the storage limit binds. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 30.678776706834583,
        "ec": 87.72652299304896,
        "etp": 175.45304598609786,
        "prcp": 0.0006315885111689568,
        "shdfac": 0.5,
        "cmc": 0.0005000000237487257,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step601_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 2.80717677022483e-07,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    4.68154583432856e-06,
                    2.877064420037101e-07,
                    9.744166584724212e-09,
                    3.3047509173655953e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": 4.68154583432856e-06,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": 2.877064420037101e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": 9.744166584724212e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.3047509173655953e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    4.68154583432856e-06,
                    2.877064420037101e-07,
                    9.744166584724212e-09,
                    3.3047509173655953e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step639_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at shallow convection, light rain with the storage limit just binding. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 27.87296419109178,
        "ec": 78.55144999859444,
        "etp": 158.27589909302816,
        "prcp": 1.9605622583185323e-05,
        "shdfac": 0.5,
        "cmc": 0.0004926163819618523,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step639_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.165972464551076e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step639_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at shallow convection, light rain with the storage limit just binding. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 27.87296419109178,
        "ec": 78.55144999859444,
        "etp": 158.27589909302816,
        "prcp": 1.9605622583185323e-05,
        "shdfac": 0.5,
        "cmc": 0.0004926163819618523,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step639_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -2.1605205504423436e-08,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -1.000612109836041e-06,
                    2.9134455403756476e-07,
                    1.2103362756477054e-08,
                    3.3266056576053415e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -1.000612109836041e-06,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": 2.9134455403756476e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": 1.2103362756477054e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.3266056576053415e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -1.000612109836041e-06,
                    2.9134455403756476e-07,
                    1.2103362756477054e-08,
                    3.3266056576053415e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step689_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at late afternoon, light rain, storage limit binding. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 23.990589199279373,
        "ec": 64.79999707320673,
        "etp": 139.3510556485384,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0004324734618421644,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step689_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.168425307287893e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step689_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at late afternoon, light rain, storage limit binding. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 23.990589199279373,
        "ec": 64.79999707320673,
        "etp": 139.3510556485384,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0004324734618421644,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step689_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -2.590963497353313e-08,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -8.319793576294643e-07,
                    2.0002838274635337e-07,
                    1.3282552835391925e-08,
                    3.3620828343572384e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -8.319793576294643e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": 2.0002838274635337e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": 1.3282552835391925e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.3620828343572384e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -8.319793576294643e-07,
                    2.0002838274635337e-07,
                    1.3282552835391925e-08,
                    3.3620828343572384e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step918_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at night, no rain. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 0.15186052072666764,
        "ec": 0.405269177921512,
        "etp": 0.9313998858967502,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.000378655968233943,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step918_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.179673125619928e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step918_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at night, no rain. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 0.15186052072666764,
        "ec": 0.405269177921512,
        "etp": 0.9313998858967502,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.000378655968233943,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step918_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -1.6204285634667526e-10,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -2.5789417268384796e-07,
                    3.3746316852045766e-08,
                    1.9111218163558874e-08,
                    3.5991154501147093e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -2.5789417268384796e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": 3.3746316852045766e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": 1.9111218163558874e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.5991154501147093e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -2.5789417268384796e-07,
                    3.3746316852045766e-08,
                    1.9111218163558874e-08,
                    3.5991154501147093e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step1080_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at deep night, no rain. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 0.05385085629674085,
        "ec": 0.14598072086361838,
        "etp": 0.33597324879085616,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0003775822988245636,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step1080_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.18803066720217e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step1080_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at deep night, no rain. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 0.05385085629674085,
        "ec": 0.14598072086361838,
        "etp": 0.33597324879085616,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0003775822988245636,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step1080_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -5.836894010659288e-11,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -1.4830981998059656e-07,
                    -2.4551083388502093e-09,
                    1.8720924810367023e-08,
                    3.786781999082223e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -1.4830981998059656e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": -2.4551083388502093e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": 1.8720924810367023e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.786781999082223e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -1.4830981998059656e-07,
                    -2.4551083388502093e-09,
                    1.8720924810367023e-08,
                    3.786781999082223e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step1621_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at second-day morning, peak evaporative demand, no rain. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 50.003827656275234,
        "ec": 90.61044303911503,
        "etp": 322.25223310635715,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.00015812285710126162,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step1621_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.218911703306421e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step1621_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at second-day morning, peak evaporative demand, no rain. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 50.003827656275234,
        "ec": 90.61044303911503,
        "etp": 322.25223310635715,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.00015812285710126162,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step1621_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -3.622968534065015e-08,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -2.38657388296204e-07,
                    -6.539865871069139e-08,
                    -1.2379614000579409e-08,
                    4.200878533922037e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -2.38657388296204e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": -6.539865871069139e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": -1.2379614000579409e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 4.200878533922037e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -2.38657388296204e-07,
                    -6.539865871069139e-08,
                    -1.2379614000579409e-08,
                    4.200878533922037e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step1730_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at second-day deep convection, rain fully absorbed. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 42.20656617467408,
        "ec": 23.896544158421932,
        "etp": 275.56436111336416,
        "prcp": 8.39278527564602e-06,
        "shdfac": 0.5,
        "cmc": 1.5040217476780526e-05,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step1730_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.225082825938179e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step1730_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at second-day deep convection, rain fully absorbed. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 42.20656617467408,
        "ec": 23.896544158421932,
        "etp": 275.56436111336416,
        "prcp": 8.39278527564602e-06,
        "shdfac": 0.5,
        "cmc": 1.5040217476780526e-05,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step1730_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -5.358403107211468e-09,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -1.4380246993184187e-07,
                    -9.031161374295493e-08,
                    -2.904163021177908e-08,
                    3.920475055707584e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -1.4380246993184187e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": -9.031161374295493e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": -2.904163021177908e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.920475055707584e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -1.4380246993184187e-07,
                    -9.031161374295493e-08,
                    -2.904163021177908e-08,
                    3.920475055707584e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step1801_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at second-day noon, rain fully absorbed. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 42.11480590968112,
        "ec": 7.740982689672649,
        "etp": 276.75870288782556,
        "prcp": 4.9646823754301295e-06,
        "shdfac": 0.5,
        "cmc": 1.5646587598894257e-06,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step1801_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.229084926905489e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step1801_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at second-day noon, rain fully absorbed. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 42.11480590968112,
        "ec": 7.740982689672649,
        "etp": 276.75870288782556,
        "prcp": 4.9646823754301295e-06,
        "shdfac": 0.5,
        "cmc": 1.5646587598894257e-06,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step1801_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -6.128138261394405e-10,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -1.5384151730302165e-07,
                    -9.822876401166525e-08,
                    -3.582554253256376e-08,
                    3.716849050761084e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -1.5384151730302165e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": -9.822876401166525e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": -3.582554253256376e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.716849050761084e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -1.5384151730302165e-07,
                    -9.822876401166525e-08,
                    -3.582554253256376e-08,
                    3.716849050761084e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step2101_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at second-day afternoon, rain heavy enough for surface runoff with the canopy NOT full. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 26.538493717737392,
        "ec": 81.07784972891477,
        "etp": 162.15569945782954,
        "prcp": 3.429367279750295e-05,
        "shdfac": 0.5,
        "cmc": 0.0005000000237487257,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step2101_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 1.716308928468924e-11,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.243004560710315e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step2101_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at second-day afternoon, rain heavy enough for surface runoff with the canopy NOT full. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 26.538493717737392,
        "ec": 81.07784972891477,
        "etp": 162.15569945782954,
        "prcp": 3.429367279750295e-05,
        "shdfac": 0.5,
        "cmc": 0.0005000000237487257,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step2101_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -1.5271336224666893e-08,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -3.5300403156135474e-07,
                    1.0787443160964472e-07,
                    9.706341286275233e-09,
                    3.10848569107236e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -3.5300403156135474e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": 1.0787443160964472e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": 9.706341286275233e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.10848569107236e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -3.5300403156135474e-07,
                    1.0787443160964472e-07,
                    9.706341286275233e-09,
                    3.10848569107236e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step2134_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at second-day shallow convection, storage limit just binding. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 20.956293814792314,
        "ec": 63.73840730703794,
        "etp": 128.46640000139237,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0004923266242258251,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step2134_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.244411200401144e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step2134_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at second-day shallow convection, storage limit just binding. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 20.956293814792314,
        "ec": 63.73840730703794,
        "etp": 128.46640000139237,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0004923266242258251,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step2134_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -2.548516885806512e-08,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -4.704275524680668e-07,
                    9.577841297847556e-08,
                    1.0171197217800909e-08,
                    3.1252667120895694e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -4.704275524680668e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": 9.577841297847556e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": 1.0171197217800909e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.1252667120895694e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -4.704275524680668e-07,
                    9.577841297847556e-08,
                    1.0171197217800909e-08,
                    3.1252667120895694e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step2400_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at second-day night, no rain. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 0.061265701433643405,
        "ec": 0.17719912716467395,
        "etp": 0.3890426222330375,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0004149146843701601,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step2400_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.256431608381364e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step2400_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at second-day night, no rain. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 0.061265701433643405,
        "ec": 0.17719912716467395,
        "etp": 0.3890426222330375,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0004149146843701601,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step2400_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -7.085131092932073e-11,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -1.4202615195024748e-07,
                    1.0438783171196064e-08,
                    1.2043088748470154e-08,
                    3.288896932573948e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -1.4202615195024748e-07,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": 1.0438783171196064e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": 1.2043088748470154e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.288896932573948e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -1.4202615195024748e-07,
                    1.0438783171196064e-08,
                    1.2043088748470154e-08,
                    3.288896932573948e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step2521_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. INFILTRATION AND RUNOFF at third-day midnight, the DEW step: etp < 0, so the only water input is dew, routed through the canopy exactly like rain. Reference at ESM_DT = 1e-4 s, which is NOT the smallest available and is chosen deliberately. WRF's infiltration limiter contains val = 1 - exp(-kdt dt/86400); as dt falls the argument underflows into the cancellation of 1 - exp(-x), and as dt rises val departs from its linear limit, so the replay's own accuracy is a V in dt. Measured on the step-601 column, the replay differs from this component by 5.8e-7 at dt = 1e-6, 1.1e-8 at dt = 1e-4 and 2.2e-5 at dt = 1, with a clean minimum at 1e-4; the runoff1 tolerance below is set from that floor and bounds the REFERENCE, not the component. runoff2 contains no dt-dependent quantity at all and is pinned at 1e-9.",
      "parameter_overrides": {
        "edir": 0.014123730433023923,
        "ec": 0.04115429419927649,
        "etp": 0.0903939790520061,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.000414554204326123,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_inf_step2521_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-18
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 5.262465638588747e-10,
          "tolerance": {
            "rel": 1e-09
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "chem170_scm_step2521_tendency",
      "description": "Reference: WRF 4.8 em_scm_xy single-column run at chem_opt = 170 with Noah (sf_surface_physics = 2, num_soil_layers = 4, silt loam, dryland cropland, slopetyp 1), data/eqwefic/chem170_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-chem170 and replayed through the real64 (-fdefault-real-8) kernels/noah_driver. Inputs are the dumped SFLX arguments (real32 values widened). The column is unfrozen and unsaturated at every dumped step, so the frozen-ground infiltration reduction fcr and SSTEP's supersaturation redistribution are UNTESTED, not merely unused, and so is every path that reads sice. SOIL WATER AND CANOPY TENDENCIES at third-day midnight, the DEW step: etp < 0, so the only water input is dew, routed through the canopy exactly like rain. SSTEP advances the soil column IN PLACE with an implicit tridiagonal solve, so WRF's dt = 60 s increment is not the instantaneous tendency; the reference is the replay increment at ESM_DT = 0.01 s. drip and dcmc_dt are referenced at the SAME step and not at a smaller one, because WRF forms drip as (cmc + dt rhsct) - cmcmax: the increment is nine orders below cmc at dt = 1e-6 and is destroyed by cancellation, while at 1e-2 it is recovered to 1e-11. The per-level assertions carry an ABSOLUTE bound: the reference's own implicit-solve truncation is about 1e-11 (1/s), which is a few times 1e-5 of the DEEPEST layer's tendency, so the quiet layers cannot be pinned relatively at this step size while the top layer is pinned to 1e-7 of its own amplitude.",
      "parameter_overrides": {
        "edir": 0.014123730433023923,
        "ec": 0.04115429419927649,
        "etp": 0.0903939790520061,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.000414554204326123,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_tend_step2521_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "drip",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 1e-09,
            "abs": 1e-18
          }
        },
        {
          "variable": "dcmc_dt",
          "time": 0.0,
          "expected": -1.645513694921441e-11,
          "tolerance": {
            "rel": 1e-06,
            "abs": 1e-16
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -9.922910271242813e-08,
                    -3.3669067533992347e-09,
                    1.1721368320394276e-08,
                    3.366823486672388e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 1
          },
          "expected": -9.922910271242813e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 2
          },
          "expected": -3.3669067533992347e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 3
          },
          "expected": 1.1721368320394276e-08,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsh2o_dt",
          "time": 0.0,
          "coords": {
            "lev": 4
          },
          "expected": 3.366823486672388e-09,
          "tolerance": {
            "rel": 0.001,
            "abs": 5e-11
          }
        },
        {
          "variable": "dsmc_dt",
          "time": 0.0,
          "reduce": "Linf_error",
          "expected": 0.0,
          "reference": {
            "op": "faq",
            "output_idx": [
              "k"
            ],
            "args": [],
            "ranges": {
              "k": {
                "from": "lev"
              }
            },
            "expr": {
              "op": "index",
              "args": [
                {
                  "op": "const",
                  "args": [],
                  "value": [
                    -9.922910271242813e-08,
                    -3.3669067533992347e-09,
                    1.1721368320394276e-08,
                    3.366823486672388e-09
                  ]
                },
                "k"
              ]
            }
          },
          "tolerance": {
            "abs": 1e-10,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_A1_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case A step 1: sicemax = 0.1435, dice = 0.1895 m, ABOVE the 1e-2 m threshold, so the frozen-ground reduction fcr is ACTIVE here -- the only step in the whole reference set that reaches it.",
      "parameter_overrides": {
        "edir": 0.0,
        "ec": 0.0,
        "etp": -18.454618195736202,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_A1_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 1.764719749591192e-09,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 1.6090999396490623e-11,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1000_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1000: sicemax = 0.0281, dice = 0.0040 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 0.0,
        "ec": 0.0,
        "etp": -1.0102365010923242,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 3.7311679079721216e-06,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1000_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 1.5141246595384566e-15,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1025_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1025: sicemax = 0.0295, dice = 0.0042 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 0.0,
        "ec": 0.0,
        "etp": -1.033531198835705,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 4.037526196043473e-06,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1025_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 1.5847576682150931e-15,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1501_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1501: sicemax = 0.0478, dice = 0.0061 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 0.0,
        "ec": 0.0,
        "etp": -1.5126425258350573,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 1.132428769778926e-05,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1501_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 3.3946301933216973e-15,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1526_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1526: sicemax = 0.0484, dice = 0.0062 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 0.0,
        "ec": 0.0,
        "etp": -1.5348410017577732,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 1.1781132343458012e-05,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1526_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 3.4949979308154492e-15,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1564_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1564: sicemax = 0.0485, dice = 0.0063 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 0.017098857547683834,
        "ec": 0.0308665075703282,
        "etp": 0.39551551933085194,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 1.2180859812360723e-05,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1564_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1787_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1787: sicemax = 0.0174, dice = 0.0033 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 11.200020493405772,
        "ec": 0.0,
        "etp": 260.7424722140067,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1787_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1813_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1813: sicemax = 0.0134, dice = 0.0029 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 11.659375142523958,
        "ec": 0.0,
        "etp": 271.9437095916205,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1813_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1838_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1838: sicemax = 0.0092, dice = 0.0025 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 11.849831023237003,
        "ec": 0.0,
        "etp": 276.9302750513138,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1838_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1928_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1928: sicemax = 0.0054, dice = 0.0016 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 10.615661132036081,
        "ec": 0.0,
        "etp": 250.05675816487394,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1928_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1934_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1934: sicemax = 0.0054, dice = 0.0016 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 10.435622247015665,
        "ec": 0.0,
        "etp": 245.94110192691295,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1934_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1935_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1935: sicemax = 0.0054, dice = 0.0016 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 10.404153994925844,
        "ec": 0.0,
        "etp": 245.22008186638493,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1935_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D1953_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 1953: sicemax = 0.0054, dice = 0.0016 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 9.77735110216749,
        "ec": 0.0,
        "etp": 230.78501140568616,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D1953_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D2096_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 2096: sicemax = 0.0049, dice = 0.0015 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 0.06917542825644829,
        "ec": 0.0,
        "etp": 1.6429748818858085,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 0.0,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D2096_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D2144_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 2144: sicemax = 0.0046, dice = 0.0014 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 0.0,
        "ec": 0.0,
        "etp": -0.7701807617128119,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 1.1624429419043736e-07,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D2144_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 8.78660165818619e-16,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D2169_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 2169: sicemax = 0.0045, dice = 0.0016 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 0.0,
        "ec": 0.0,
        "etp": -0.7415429794609039,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 3.472641481039318e-07,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D2169_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 8.14532778468076e-16,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    },
    {
      "id": "cold_D2880_infiltration",
      "description": "Reference: WRF 4.8 em_scm_xy COLD-SEASON single-column run, data/eqwefic/noahcold_scm, dumped by ctessum-claude/WRF branch earthsciml-instrumented-noahcold and replayed through the real64 (-fdefault-real-8) kernels/noah_driver at ESM_DT = 1e-6 s. Same soil and vegetation as the warm run (silt loam, dryland cropland), so only the cold physics differs. These are the NOPAC steps -- bare frozen ground with no snow pack -- which is the regime this component describes; the SNOPAC steps are NOT used, because there the water reaching the soil is snowmelt routed through the pack rather than the prcp this component reads. WHAT THESE ADD over the warm tests: soil ice is present at every one of them, so the paths that read sice -- sicemax, the vkwgt ice weighting of the diffusivity, the sice term in the storage deficit, and the frozen-ground infiltration reduction fcr -- are exercised here for the first time. There is NO precipitation at any of these steps: the water supply is DEW, which NOPAC adds to prcp1 before the canopy split, so these tests also pin the dew routing on a second, independent set of columns. The soil-water TENDENCY is deliberately NOT asserted here. On frozen ground SFLX's SHFLX applies the phase-change sink SNKSRC to sh2o AFTER SMFLX has moved it, so (sh2o_out - sh2o_in)/dt is the sum of transport and freezing, and this component supplies only the transport; the frozen sink is a separate component that does not yet exist (FORTRAN_BUGS N95). Everything asserted below is computed in SMFLX BEFORE SHFLX runs and is therefore unaffected by it. Case D step 2880: sicemax = 0.0379, dice = 0.0048 m, below the 1e-2 m threshold, so fcr is exactly 1 (its inactive branch).",
      "parameter_overrides": {
        "edir": 0.0,
        "ec": 0.0,
        "etp": -1.0021067894151603,
        "prcp": 0.0,
        "shdfac": 0.5,
        "cmc": 7.408653800666798e-06,
        "smcmax": 0.47600001096725464,
        "smcwlt": 0.08399999886751175,
        "bexp": 5.329999923706055,
        "dksat": 2.8100000690756133e-06,
        "dwsat": 2.390000008745119e-05,
        "kdt": 4.215000114255272,
        "frzx": 0.17460114362402857,
        "slope": 0.10000000149011612
      },
      "expression_template_imports": [
        {
          "ref": "./tests/moist_cold_D2880_inputs.esm"
        }
      ],
      "time_span": {
        "start": 0.0,
        "end": 1.0
      },
      "assertions": [
        {
          "variable": "runoff1",
          "time": 0.0,
          "expected": 1.4875400338444707e-15,
          "tolerance": {
            "rel": 5e-08,
            "abs": 1e-20
          }
        },
        {
          "variable": "runoff2",
          "time": 0.0,
          "expected": 2.0168126413987087e-12,
          "tolerance": {
            "rel": 1e-08
          }
        },
        {
          "variable": "supersaturation_active",
          "time": 0.0,
          "expected": 0.0,
          "tolerance": {
            "abs": 0.0,
            "rel": 0.0
          }
        }
      ]
    }
  ]
}