Multiscale mapping of burn area and severity using multisensor satellite data and spatial autocorrelation analysis

Traditional methods of recording fire burned areas and fire severity involve expensive and time-consuming field surveys. Available remote sensing technologies may allow us to develop standardized burn-severity maps for evaluating fire effects and addressing post fire management activities. This paper focuses on multiscale characterization of fire severity using multisensor satellite data. To this aim, both MODIS (Moderate Resolution Imaging Spectroradiometer) and ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) data have been processed using geo-statistic analyses to capture pattern features of burned areas.Even if in last decades different authors tried to integrate geo-statistics and remote sensing image processing, methods used since now are only variograms, semivariograms and kriging. In this paper, we propose an approach based on the use of spatial indicators of global and local autocorrelation. Spatial autocorrelation statistics, such as Moran's I and Getis–Ord Local Gi index, were used to measure and analyze dependency degree among spectral features of burned areas. This approach enables the characterization of pattern features of a burned area and improves the estimation of fire severity.

► Use of spatial indicators of autocorrelation to detect burned areas.
► Use of spatial indicators of autocorrelation to estimate fire severity.
► Multiscale analysis using both MODIS and ASTER data to identify burned areas.
► Approach independent on sensors used and on vegetation cover types.
► Burn severity is important for monitoring post-fire dynamic and vegetation resilience.