Suitability of Reanalysis products in capturing Flood Inundation and Hazards over India: Deriving insights through Statistical tests and Numerical Flood Modeling

India is infamous for the highest proportion of its population that is exposed to direct and indirect flood impacts. Despite disaster looming situations, several flood-prone basins in the country still lack adequate ground-based coverage of gauge stations; thus, hindering our comprehension of flood impacts via numerical flood modelling. Reanalysis datasets, an advancement from Earth Observation Datasets (EOD), emerge as a valuable substitute for sparse ground-based observations as they furnish relevant atmospheric and hydrological variables at high spatio-temporal resolutions. This study evaluates the efficacy of runoff and rainfall estimates from high-resolution ERA-5, JRA-55, CFSR, and MERRA Hydrological Reanalysis Data (HRD) across India for capturing flood inundation and hazards. The runoffs and rainfalls in each Reanalysis dataset are compared with the ground-based observations using various metrics such as correlation coefficient (CC) and Kling-Gupta efficiency (KGE). In the next step, they are considered as primary boundary conditions along with other ancillary datasets to LISFLOOD-FP, a global hydrodynamic flood model, to derive high-resolution flood maps for specific flood events. The simulated flood inundation maps are calibrated and validated against past flood incidences derived from satellite altimetry using performance indices including Hit-Rate (HR), False Alarm Ratio (FAR), Error Bias (EB), and Critical Success Index (CSI).  Subsequently, the best-performing HRD for a specific basin is utilized to derive distributed design input values through extreme value analysis for various scenarios (e.g., 1 in 50-yr, 100-yr, and 200-yr). These distributed discharges are fed to LISFLOOD-FP to generate high-resolution flood inundation and hazard maps. The study, for the first time, determines the efficacy of Reanalysis products in flood mapping over data-limiting large watersheds, thus providing a solid foundation for stepping up for quantifying flood risks, even under changing climate conditions.