A MODIS direct broadcast algorithm for mapping wildfire burned area in the western United States

摘要: Abstract Improved wildland fire emission inventory methods are needed to support air quality forecasting and guide the development of shed management strategies. Air requires dynamic estimates that generated in a timely manner real-time operations. In regulatory planning realm, inventories essential for quantitatively assessing contribution wildfire pollution. The depends on burned area as critical input. This study presents Moderate Resolution Imaging Spectroradiometer (MODIS) – direct broadcast (DB) mapping algorithm designed development. combines active locations single satellite scene burn scar detections provide rapid yet robust area. Using U.S. Forest Service Fire Sciences Laboratory (FiSL) MODIS-DB receiving station Missoula, Montana, provided daily measurements events western 2006 2007. We evaluated algorithm's detection rate using perimeter data information derived from high resolution imagery. FiSL system detected 87% all reference fires > 4 km 2 , 93% > 10 km . was highly correlated ( R  = 0.93) with imagery dataset, but exhibited large over estimation (56%). dataset used calibrate response quantify uncertainty measurement at incident level. An objective, empirical error based approach employed our metric is meaningful context remotely sensed inventories. ± 36% 50 km size, improving ± 31% size 100 km Fires this range account substantial portion (77% due > 50 km 66% results > 100 km ). dominance these wildfires area, duration, emissions makes significant concern forecasters regulators. With coverage 1-km spatial resolution, quantified uncertainty, presented paper well suited Furthermore, DB implementation enables time sensitive operational forecasting.