A comparative evaluation of MODIS/ASTER airborne simulator (MASTER) data and burn indices for mapping southern California fires

Large fires occur annually in southern California, producing impacts at a number of scales, from local impacts on vegetation, hydrology and microclimates, to global impacts such as emissions, affecting atmospheric chemistry, air quality, radiation balance and biogeochemical cycling. As a consequence fires are routinely mapped using various sensors and burn indices. However, the indices employed for mapping these fires have not been developed and optimized for mapping southern California burned surfaces. Therefore, this study utilizes the high spatial and spectral resolution imagery from the MODIS/ASTER airborne simulator (MASTER) to identify the most effective bands and indices specifically for burned area mapping of the southern California region. The fire perimeter is based on the Burned Area Reflectance Classification (BARC) map created by the United States Forest Service (USFS), Remote Sensing Applications Center (RSAC) and a supervised classification which defines the burned and unburned regions. A separability index is employed to identify the bands and indices that can best distinguish between classes. The results identify a range of well performing indices, such as the Normalized Burn Ratio (NBR) and the Vegetation Index based on mid-infrared spectral region (VI3), and some poor performing indices, such as the Global Environment Monitoring Index (GEMI) and the Burned Area Index (BAI). Additionally this study highlights the indices that perform better over certain vegetation types. These results are useful for understanding the application of remotely sensed data for mapping burned surfaces. Improved burned area mapping capabilities are essential for informing land managers when identifying regions susceptible to hazards (such as debris and flood flows) and for deciding where to allocate time and resources in recovery efforts. Additionally, these results can be used to validate other sensors that are used to map burned surfaces on greater spatial and temporal scales.