Remote sensing delineation of wildfire spatial extents and post-fire recovery along a semi-arid climate gradient

Understanding the wildfire extent and post-fire vegetation recovery is critical for fire and forest management. Remote sensing imagery is widely used in wildfire detection because it provides continuous and large-scale surface monitoring capability. In this study, we apply and evaluate the performance of the LandTrendr algorithm in wildfire detection across a semi-arid climate region with a marked precipitation gradient. The aims are to compare the performance of four spectral indices for the burned areas and post-fire recovery detection for different climate conditions and investigate the relationship between suitable model parameters and climate conditions. The results show that NDVI outperforms other indices, including NBR, in burned area detection for drier areas (annual precipitation <400 mm). Disturbance signal-to-noise ratio can serve as an indicator for suitable index selection for semi-arid areas. Although the performance in the detection of burned pixels varies among different indices, they are all suitable for delineating post-fire recovery except for the wet site (annual precipitation of 575 mm) where NBR displays the best performance. Parameter optimization results along the climate gradient show that climate conditions have a significant impact on suitable parameter selection. These findings provide guidance on vegetation wildfire detection in arid and semi-arid climates to support wildfire risk and forestry management.

Automated summary

Understanding wildfire extent and post-fire vegetation recovery is crucial for fire and forest management. This study evaluates the performance of the LandTrendr algorithm in wildfire detection in a semi-arid climate region and compares different spectral indices for burned area detection and post-fire recovery. The results show that NDVI performs better than other indices in detecting burned areas in drier regions.