Data for "Declines in conifer recovery and forest loss from four decades of wildfire in California"

Bhoot, V.N., Kim, J.E., Randerson, J.T., Goulden, M.L., 2026. Declines in Conifer Forest Recovery and Forest Loss From Four Decades of Wildfire in California. JGR Biogeosciences 131, e2025JG009105. https://doi.org/10.1029/2025JG009105 We used a random forest model with LANDFIRE's 2022 Existing Vegetation Type as the reference to produce EVT for years 1985-2022 (noted as RF-EVT here). R, an open source programming langauge, was used for production of RF-EVT. Original LANDFIRE 2022 Existing Vegetation Type reference layer used for production of our layers are available from https://www.landfire.gov/data/FullExtentDownloads. Further methodology for raster stack produced in the analysis are detailed in manuscript methods section. Also incldued in the repository are analysis ready data tables and R scripts for manuscript results. Analysis was done starting with physiognomic groupings of RF-EVT, column "EVT_PHYS" provided in "LF22_EVT_230.csv" in the "Bhootetal_EVT_code_data_figures.zip" file under the "Data" folder. We aggregated these for simplification of the analysis, aggregations provided in "EVT_PHYS_Code_with_coarse.csv" in the same location (referred to as EVTP in the text). For users of the rasters, the finer the classification the more noisey the data is expected to be, careful consideration must be taken if using RF-EVT over coarser cross-walks of RF-EVT (e.g. physiognomic groups). To retain the LANDFIRE color table when plotting in R, download respective ".tif.aux.xml" file for the raster(s). "LF22_EVT_230.csv" also contains the respective RGB for each RF-EVT as well.