Wildfire

Statistical fire containment probability model

Logistic regression model for large fire occurrence probability

Pure-I/O data loader for MTBS (Monitoring Trends in Burn Severity) post-fire burn-severity products. Provides the time-invariant final burn-severity raster for a single fire: a thematic severity class layer (1=unburned/low … 6=non-processing-area mask) and the continuous dNBR (differenced Normalized Burn Ratio) layer, on the native Albers/geographic raster the MTBS product ships. Sourced as GeoTIFF bands of the per-fire MTBS bundle (one final product per fire — there is no temporal progression).

Pure-I/O data loader for NIFC / GeoMAC daily wildfire perimeters, RASTERIZED OFFLINE to a burned-fraction grid. Provides a daily burned_fraction(lon,lat) field in [0,1] — the per-cell fraction of area inside the cumulative (or daily) fire perimeter — read as an ordinary gridded time series (one NetCDF per day). This is the day-by-day fire PROGRESSION reference for fire-model validation (campfire-e2e Phase E1/E2).

Wildfire plume-rise model based on Sofiev et al. (2012). Computes plume top height H_p (m) as a function of atmospheric boundary-layer height H_abl (m), free-troposphere Brunt-Väisälä frequency N_ft (s⁻¹), and fire radiative power P_fr (W) via one empirical algebraic equation: H_p = α·H_abl + β·(P_fr/P_f0)^γ · exp(−δ·N_ft²/N_0²). The five empirical constants (α=0.24, β=170 m, γ=0.35, δ=0.6, N_0=√(2.5×10⁻⁴) s⁻¹) are from Sofiev et al. (2012). In a coupled model H_abl and N_ft are supplied by GEOSFP reanalysis; the additional variable lev_p (plume-top pressure level index) is computed from H_p by the GEOSFP coupler via a Newton-iteration height-to-level mapping. Migrated from EnvironmentalTransport.jl@a48559c4114eb9f06668ecf9f06235aad20f63f0 src/plume_rise/sofiev_2012.jl:14.

Pure-I/O data loader for NASA FIRMS VIIRS active-fire detections. Provides per-overpass point detections (one row per fire pixel per satellite pass): fire radiative power, brightness temperature, and detection confidence at irregular (lon,lat) detection locations resolved at runtime. This is the SUB-DAILY fire-front timing reference for fire-model validation (campfire-e2e Phase E1/E2) — the best progression signal between daily perimeters.