Beyond Landsat: a comparison of four satellite sensors for detecting burn severity in ponderosa pine forests of the Gila Wilderness, NM, USA

Methods of remotely measuring burn severity are needed to evaluate the ecological and environmental impacts of large, remote wildland fires. The challenges that were associated with the Landsat program highlight the need to evaluate alternative sensors for characterising post-fire effects. We compared statistical correlations between 55 Composite Burn Index field plots and spectral indices from four satellite sensors varying in spatial and spectral resolution on the 2003 Dry Lakes Fire in the Gila Wilderness, NM. Where spectrally feasible, burn severity was evaluated using the differenced Enhanced Vegetation Index (dEVI), differenced Normalised Difference Vegetation Index (dNDVI) and the differenced Normalised Burn Ratio (dNBR). Both the dEVI derived from Quickbird and the dNBR derived from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) showed similar or slightly improved correlations over the dNBR derived from Landsat Thematic Mapper data (R2 = 0.82, 0.84, and 0.78 respectively). The relatively coarse resolution MODIS-derived NDVI image was weakly correlated with ground data (R2 = 0.38). Our results suggest that moderately high-resolution satellite sensors like Quickbird and ASTER have potential for providing accurate information about burn severity. Future research should develop stronger links between higher-resolution satellite data and burn severity across a range of environments.