MODIS-Landsat data fusion for continental scale burned area mapping

Satellite data have been used to monitor fire for more than two decades using computer algorithms that detect the location of active fires at the time of satellite overpass, and in the last decade using burned area mapping algorithms that map the spatial extent of the areas affected by fires. Until the successful launch of the polar-orbiting NASA Moderate Resolution Imaging Spectroradiometer (MODIS) sensors there were no environmental satellite systems with dedicated fire monitoring capabilities. The MODIS design includes bands specifically selected for fire detection and MODIS data are being used to systematically generate the daily global 1km active fire and the monthly 500m burned area products. However, neither MODIS product can detect fires reliably at the scale of 10's of meters. The recent U.S. free Landsat data policy now provides the opportunity for continental to global scale Landsat 30m resolution processing. This paper presents a multi-temporal methodology to fuse the MODIS active fire and burned area products with Landsat data to map burned areas at 30m on a temporally rolling basis. To demonstrate the fusion methodology, 30m burned area maps of the conterminous United States (CONUS) are generated using the freely available Web Enabled Landsat (WELD) ETM+ mosaics (http://landsat.usgs.gov/WELD.php). Validation is conducted by systematic comparison with the fire perimeter vectors provided by the USGS Monitoring Trends in Burn Severity project. Prospects for future developments and continental application are discussed. The presented methodology demonstrates the potential for the fusion of the planned NPP/NPOESS VIIRS active fire product with reflectance data sensed by the planned Landsat Data Continuity missions.