Mapping wildfire damage and post-fire recovery in the Westlake Wetlands, South Africa, using Sentinel-2 time series

Wildfires are becoming increasingly frequent and severe under changing climatic conditions, creating growing threats to sensitive wetland ecosystems. Despite their ecological importance, little information exists on post-fire vegetation recovery in urban wetlands of the Cape Peninsula. This study aimed to quantify the spatial extent of wildfire damage and to monitor the short-term vegetation recovery in the Westlake Wetlands following the December 2024 wildfire using Sentinel-2 imagery. Google Earth Engine and Sentinel-2 imagery was used to perform Normalised Burn ratio (NBR) and Normalised Differenced Vegetation Index (NDVI) before the fire and 2 days after the fire, with the results being used to perform Differenced NBR and differenced NDVI. The differenced indices evaluated the burn severity, as well as the condition of the vegetation due to the wildfire, respectively. Post-fire recovery assessment was conducted through the application of NBR and NDVI 10 months post-fire, in alignment with suitable imagery that was available during the study period. These were compared with the immediate post-fire damage indices that were conducted 2 days post-fire through dNDVI and dNBR to assess the extent of burn severity of recovery of the burned area and the vegetation conditions. The burn severity was interpreted using the United States Geological Survey(USGS) suggested burn severity classification table, while vegetation recovery was interpreted according to remote sensing classification values that are commonly used and accepted for NDVI interpretation. PlanetScope imagery was used as higher-resolution contextual evidence to visually corroborate the spatial patterns identified from Sentinel-2 imagery. ArcGIS Pro was used for visualisation and spatial analysis. Results showed that approximately 9% of the wetland experienced high burn severity. In the immediate post-fire to ten-month comparison, spectral greenness had increased from approximately 3% to 29%, and more than 80% of the burnt area returned to healthy NDVI values indicating natural regeneration. These findings demonstrate substantial short-term recovery in vegetation greenness within the Westlake Wetlands following a moderate wildfire and highlight the effectiveness of freely available Sentinel-2 imagery and cloud-based remote sensing workflows for monitoring post-fire ecosystem recovery. The approach provides a cost-effective framework for supporting wetland conservation, wildfire management and long-term ecological monitoring in fire-prone Mediterranean-type ecosystems.