Flood forecasting in a trans-boundary river basin is challenging due to insufficient data sharing between countries in the upper and lower reaches of a basin. A solution is the use of satellite observed rainfall and numerical weather prediction (NWP) for hydrological forecasting. We applied this method to the trans-boundary sparsely gauged Chenab River basin in Pakistan and India to reproduce the exceptionally high flood in 2014. We employed global NWPs by three weather centers to consider forecast uncertainty and downscaled them using the Weather Research and Forecasting (WRF) model to prepare precipitation inputs. For hydrological simulations, we used a kinematic wave model, the Integrated Flood Analysis System (IFAS), for the upper-reach basin with high mountains and steep slopes, and used a diffusive-wave Rainfall-Runoff-Inundation (RRI) model for low altitudes and mild slopes. In our forecasting experiment, the precipitation by the global NWP were not able to predict flood peaks consistently. However, the downscaled rainfall by regional NWP showed good performance in predicting flood waves quantitatively, and a multi-model approach provided added value in issuing reliable warning as early as six days in advance. A confident stream flow forecasting near the border of the countries also led to reliable inundation forecasting by the RRI model in the lower-reach basin. |