Flood forecasting system for Brahmaputra river basin

Assam has always been India’s most flood prone state due to the presence of Brahmaputra river, which is very unstable in terms of its flow direction witnessing 12 major floods from 1950 to 2012. Flooding in the basin has affected around 2.75 million of people and 0.27 million hectares of agricultural land on an average causing catastrophic damage to human life and infrastructure. In this study, we analysed all the major floods across the Brahmaputra river in the past 70 years and established the dependency within discharge and atmospheric parameters. Variable Infiltration Capacity (VIC) model was set up to simulate the flow at two stations namely Yangcun, China and Bahadurabad, Bangladesh. We  used near surface meteorological data for driving land surface modelling systems from 1901 to 2016 as input parameters to the VIC model. To avoid the discontinuity of data after 2016, we used ECMWF reanalysis (ERA5) data for the period of 2016 to 2020. After obtaining the continuous simulated discharge for 120 years, we established the relationship between the observed and simulated discharge data for which the R-squared and Nash Sutcliffe coefficient values were 0.83 and 0.78 respectively. Comparing the simulated discharge with the observed extreme discharge at various locations on the river, we apply the model to address future flood situations.