Investigating the usefulness of satellite-retrieved land surface temperature (LST) in pre- and post-fire spatial analysis

In this paper, we analyse the applicability and effectiveness of satellite-retrieved Land Surface Temperature (LST) for monitoring and spatial analysis of forest fires using pre- and post-fire Landsat 8 and Aster satellite images. Firstly, LST was used for pre-fire analysis, along with other meteorological parameters to evaluate fire risk, and the results showed that the LST was highly correlated with all parameters. Besides, Google Earth Engine platform was used to evaluate forest-fire susceptibility in the region. For post-fire analysis, LST was used to monitor, analyse, and map post-fire disturbances using different image spectral transformations, and correlation analysis was performed to evaluate the strength of relationships between the parameters. High correlation coefficients were obtained between LST and selected spectral indices; and the highest correlation (0.90) was found between Difference Land Surface Temperature (dLST) – Difference Normalized Burn Ratio (dNBR). Spatial analysis done for dLST and slope showed that the post-fire LST increased in areas with a slope range of 42—55%. Then, spatial relation analysis between dLST and other indices was performed and a high correlation was found between dLST – dNBR, dLST – DI (Disturbance Index), and dLST – NDMI (Normalized Difference Moisture Index). Finally, for 4 categorical variables (namely CORINE Land Cover (CLC) map, dLST, Burn severity and Slope), the areal extent of burnt forest areas was calculated and most burnt areas were found to be related a certain degree with each category (i.e. coniferous forest class, -2 and 0.5 °C, medium–low burning intensity, 29–42% slope range, respectively).