Global 30m forest cover loss map (Hansen et al 2013) for 2001-2019 (updated to include 2020, 2021, 2022 and 2023) is disaggregated into forest loss due to fire vs. other disturbance drivers. The map matches the sample-based area estimates of forest loss due to fire ± SE for all continents except Africa. This allows producing sub-regional map-based area estimates with a measure of uncertainty. The sum of map pixels with code 4 (high certainty of forest loss due to fire) corresponds to sample area estimate minus SE, adding code 3 (medium certainty) to code 4 (high certainty) pixels results in map area matching the sample-based area estimate, and adding code 2 (low certainty) pixels to codes 3 and 4 (medium and high certainty) yields map area matching sample area estimate plus SE. Code 5 corresponds to all forest loss due to fire in Africa; code 1 corresponds to forest loss due to other (non-fire) drivers.
Dataset Reference
Tyukavina, A., Potapov, P., Hansen, M.C., Pickens, A., Stehman, S., Turubanova, S., Parker, D., Zalles, V., Lima, A., Kommareddy, I., Song, X-P, Wang, L. and Harris, N. (2022) Global trends of forest loss due to fire, 2001-2019. Frontiers in Remote Sensing https://doi.org/10.3389/frsen.2022.825190 [Open PDF]
Data visualization using Google Earth Engine Apps
https://glad.earthengine.app/view/global-forest-loss-due-to-fire
Dataset Access in Google Earth Engine
Image Collection id: users/sashatyu/2001-2023_fire_forest_loss
Data values: codes 2-4 - forest loss due to fire, low-high certainty (see description above), code 5 - all forest loss due to fire in Africa, code 1 - forest loss due to other (non-fire) drivers, code 0 - no data.
Image Collection id: users/sashatyu/2001-2023_fire_forest_loss_annual
Data values: year of loss from Hansen et al. global map (codes 1-23 correspond to 2001-2023 loss) for the forest loss due to fire map matching sample area estimate (map codes 3+4, code 5 for Africa). No data = 0.
Data Download
Map data are provided in Geographic Coordinates (Lat/Long) using the WGS84 reference system.
Data format: 8-bit unsigned LZW-compressed GeoTIFF. Pixel size is 0.00025 x 0.00025 degree. Data are aggregated into continental mosaics (see region boundaries and name abbreviations in the figure below).
Files without "annual" in the name provide fire certainty levels for 2001-2023
Data values: codes 2-4 - forest loss due to fire, low-high certainty (see description above), code 5 - all forest loss due to fire in Africa, code 1 - forest loss due to other (non-fire) drivers, code 0 - no data.
Files with "_annual" in the name provide the year of loss for medium and high certainty fire pixels
Data values: year of loss from Hansen et al. global map (codes 1-23 correspond to 2001-2023 loss) for the forest loss due to fire map matching sample area estimate (map codes 3+4, code 5 for Africa). No data = 0.
