Analysis of Extreme Flood Events Using GRACE-FO and SAR Data: A Case Study of Short-Term Flood in Thessaly, Greece
摘要
Extreme weather events have increasingly resulted in significant disasters, impacting human life and economic development. Recently, short-term floods have caused substantial damage worldwide. Such was the case of the catastrophic flood that struck the region of Thessaly, Greece, in September 2023. In this study, we investigate the application of Earth Observation satellite data – optical, SAR, and gravimetric – for mapping flood events, focusing on the catastrophic flood that occurred in Thessaly. The rationale for incorporating both EO imagery and gravimetric observations lies in their complementary strengths: while optical and SAR sensors provide detailed, high-resolution snapshots of surface hydrological changes, gravimetry captures changes in the total water mass, enabling an assessment of broader hydrological conditions that may not be visible in conventional imagery. Our analysis was divided into three main components. Firstly, we utilized post-event satellite images from the Sentinel-2 platform, filtered by area of interest (AOI) and date range, with a constraint of less than 20% cloud cover. The Normalized Difference Water Index (NDWI) was applied to identify water bodies in optical imagery, serving as a reference for subsequent flood mapping methodologies. Secondly, we employed Sentinel-1 Synthetic Aperture Radar (SAR) data from pre- and post-event acquisitions. SAR images underwent edge masking and were merged to generate two mosaics. Flooded areas were detected by identifying regions with low backscatter values, specifically where the vertical vertical polarisation (VV) band exhibited values below a threshold of −17 in post-event mosaics relative to pre-event counterparts. Visual inspection confirmed the accuracy of flood detection against Sentinel-2 imagery. Lastly, we integrated GRACE and GRACE-FO Level-2 spherical harmonic coefficients provided by the Jet Propulsion Laboratory (JPL) to derive Liquid Equivalent Water Thickness (EWT), representing water thickness per grid cell. Despite the spatial resolution limitation of monthly GRACE products, EWT effectively captured spatial characteristics during short-term flood events. Simulating water level rise involved calculating flow direction, accumulation, and depressions using SRTM DEM data. The water spread simulation incorporated EWT values, modeling water movement and accumulation across terrain. This approach considered filled depressions and flow dynamics, providing a realistic representation of flood scenarios by accounting for terrain elevation and water volume. Modeled flooded areas closely resembled the spatial extent of the actual event, although volumetric accuracy was constrained by the spatial resolution of GRACE mission data.