Spatio-Temporal Flood Impact Assessment and Crop Stress Detection Using Multi-Satellite Time-Series Data: A Case Study of Bhandara District, Maharashtra, India

Abstract Floods cause widespread damage to agricultural land and crop productivity across South Asia, with monsoonseason inundation posing a recurring threat to rain-fed agricultural economies. This study presents a comprehensive spatio-temporal flood impact assessment and crop stress detection framework for Bhandara district, Maharashtra, India, using multi-satellite time-series data during the 2019 monsoon season. Sentinel-1 Synthetic Aperture Radar (SAR) data was employed for flood detection and persistence mapping using a statistically derived change detection threshold of 4.6 dB. Sentinel-2 optical imagery was used to compute the Normalized Difference Vegetation Index (NDVI) for crop stress assessment over cropland areas extracted from ESA WorldCover 10 m 2020 v100. Rainfall data from the NASA Global Precipitation Measurement (GPM) IMERG V06 product was used to establish the meteorological drivers of the flood event. Results indicate a total flood extent of 24.3 km², with 49.3 km² of persistent flooding classified into mild (31.3 km²), moderate (12.9 km²), and severe (5.1 km²) categories. Of the 2,394 km² of cropland in the district, 13.8% exhibited measurable crop stress, with a mean NDVI drop of 0.141 observed in flood-affected areas compared to non-flooded cropland. A strong positive correlation (R = 0.935) was found between monthly rainfall and flood extent. The proposed framework demonstrates the effectiveness of multi-sensor satellite data fusion for near real-time agricultural flood impact assessment, with potential applications in disaster management and agricultural planning. Keywords: Flood mapping, crop stress detection, Sentinel-1 SAR, Sentinel-2, NDVI, Google Earth Engine, Bhandara district, change detection, spatio-temporal analysis, Maharashtra, GPM IMERG, ESA WorldCover.