Detection of burned areas from mega-fires using daily and historical MODIS surface reflectance

摘要: The detection and mapping of burned areas from wildland fires is one the most important approaches for evaluating impacts fire events. In this study, a novel area algorithm rapid response applications using Moderate Resolution Imaging Spectroradiometer MODIS 500 m surface reflectance data was developed. Spectra bands 5 6, composite indices Normalized Burn Ratio, Difference Vegetation Index were employed as indicators to discover pixels. Historical statistical used provide pre-fire baseline information. Differences in current post-fire historical input into support vector machine classifier, fire-affected pixels detected mapped by classification process. Compared with existing level 3 monthly product MCD45, new able generate maps on daily basis when become available, which more applicable scenarios major incidents occur. tested three mega-fire cases that occurred continental USA. experimental results validated against perimeter database generated Geospatial Multi-Agency Coordination Group compared MCD45 product. validation indicated effective detecting caused mega-fires.