A Modified Approach for Mapping Forest Fire Severity Using Sentinel-2 Time Series in the Indian Himalayan Region

Abstract

Forest fire is an important phenomenon that influences forest ecosystems. Mapping forest fire and its severity is important for forest fire management and the restoration of fire-affected areas. This study was implemented to map forest-fire-affected areas under different fire-severity classes over a six-year period (2017-2022) in a Himalayan moist temperate forest-dominated region of the Indian Western Himalaya, which faces frequent forest fires. Identifying burnt areas is crucial for understanding various ecological processes associated with forest fire. We mapped forest fire severity using remotely sensed images. Sentinel-2 data were used to obtain the differenced normalised burn ratio (dNBR) to map burnt areas during 2017-2022, calculated as the difference between pre- and post-fire normalised burn ratio values. A protocol was developed to identify areas severely affected by fire over a long observation period. The temporal dNBR maps were classified into different fire-severity classes and matched with the percentage of active fire points detected in the burnt class by the Moderate Resolution Imaging Spectroradiometer (MODIS) and Visible Infrared Imaging Radiometer Suite (VIIRS) to validate the classification. We observed that 74.4 to 92.5% of active fire points fell within the burnt severity classes during different assessment years. The developed protocol will be helpful in identifying the spatial representation of burnt severity categories in forests, understanding various ecological processes and implementing fire-mitigation plans.