Land Cover and Vegetation Change Assessment after Extreme Precipitation Events: The Case of Tropical Storm Hanna (2020) in Northeastern Mexico Using Sentinel-2 and Random Forest

Abstract. This study evaluates land cover responses to extreme rainfall in the Monterrey Metropolitan Area (Mexico) using Sentinel-2 imagery and remote sensing techniques. Two land cover maps were generated through Random Forest classification with stratified random sampling, considering six classes: dense vegetation, medium vegetation, sparse vegetation, built-up area, bare soil, and water bodies. Three spectral indices were incorporated: the Normalized Difference Vegetation Index (NDVI), the Soil Adjusted Vegetation Index (SAVI), and the Normalized Difference Water Index (NDWI). The pre- and post-event models achieved overall accuracies of 0.93 and 0.83, with Kappa coefficients of 0.91 and 0.79, respectively. Integration with a multi-index Change Vector Analysis (CVA) enabled the detection of both categorical and magnitude-based changes, revealing significant vegetation disturbance at higher elevations, vegetation recovery in mid-elevation zones, and increased surface moisture along riparian corridors after Tropical Storm Hanna (2020). These findings demonstrate the dual effects of extreme precipitation in semi-arid mountainous urban regions and highlight the value of combining vegetation and water indices for short-term change detection. The proposed methodology is transferable to other hazard prone mountain cities worldwide, supporting disaster risk reduction, urban planning, and environmental monitoring. Limitations include the reliance on only two temporal scenes and spectral confusion between vegetation classes, which future studies could address through multi-temporal datasets, higher-resolution imagery, and integration of ancillary environmental data.