Automatic Mapping of Forest Fire and Logging Disturbances in the Western United States
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 500m burned area products. The moderate/coarse spatial resolution of the MODIS fire products is a limiting factor in the mapping accuracy of those products. Improved mapping accuracy
is needed for several fire applications including support of scientists studying pyrogenic emissions of greenhouse gasses, aerosols, and carbon budgets (http://www.gofc-gold.uni-jena.de/redd/), federal and state agencies who undertake fire management and post-fire remediation (http://earthobservatory.nasa.gov/Features/BAER/baer2.php) and for environmental management applications. Higher spatial resolution polar-orbiting satellite systems such as Landsat provide the opportunity for detailed spatial mapping of burned areas, but at a cost of reduced temporal resolution. To date, high spatial resolution satellite data have not been used for systematic burned area mapping at landscape to continental scale. This is primarily because high spatial resolution data costs have precluded large area mapping; arguably, studies have been based on what data researchers could afford and not on what they
truly needed. Since January 2008, the USGS has been providing free terrain corrected and radiometrically calibrated Landsat data via the. This revolutionary data policy provides the opportunity to use all the data in the U.S. Landsat Data Archive and to consider the systematic utility of Landsat data for long-term large-area monitoring. The NASA funded WELD project, is systematically generating 30m composited Landsat ETM+ mosaics of the CONUS and Alaska from 2006 to 2013. The WELD mosaics provide the first instance of continental science-quality Landsat data with a level of pre-processing comparable to the MODIS land products (http://landsat.usgs.gov/WELD.php). Under current NASA funding, the proposer is investigating the fusion of of MODIS fire products with WELD 30m Landsat mosaics to produce multi-temporal 30m burned area maps for the conterminous United States that will be updated on
a weekly rolling basis. Based on the preliminary results of this approach, the present proposal seeks to: (1) customize the algorithm for mapping burned areas in forest ecosystems, with a particular focus on the Western United states. (2) assemble a specific validation dataset for forest areas, based on the interpretation of high resolution satellite imagery (3) adopt the same systematic, automatic approach for monitoring industrial forestry logging in the same study area.