Evaluation of Global Water Resources Reanalysis Data for Estimating Flood Events in the Brahmaputra River Basin

Bangladesh and India are in a long-standing conflict with regard to the sharing of hydro-meteorological information of the Brahamaputra River. Consequently, it limits flood risk management in Bangladesh (the downstream country). Recently developed water resources reanalysis data appear as a promising alternative in providing this information. Further evaluation of these global datasets is needed to understand their capabilities to improve flood events estimation. In this study, the potential of the global water resources reanalysis (WRR) developed within the EU-FP7 project eartH2Observe is critically assessed for detecting and estimating flood events in the Brahmaputra River basin for 1980 to 2012 period. The discharge time series of five large-scale models available in the WRR dataset and two multi-model combinations are evaluated at different temporal resolutions and their performance is compared with a local-scale hydrological model. In situ data and reported damaging flood events compiled from two global disaster databases are used as benchmarks in flood events evaluation. Results show that the WRR data have reasonable skill in detecting flood events, though a significant underestimation of magnitude is found. This study also reveals that the individual large-scale models simulate peak flows similarly or even better than the local-scale model, capturing the hydrological behaviour in the basin and identifying the occurrence and severity of both observed and reported damaging flood events. In conclusion, this study gives insights in the applicability of global hydrological models and datasets for estimating flood events at a local-scale for transboundary rivers in water-sharing countries.