Impact of Observational Uncertainty on the Simulation of Daily Discharge of Brahmaputra River using a Multimodel Ensemble

Abstract:

The forcing given to hydrologic models influences the spatial and temporal variability of simulated streamflow. Observational uncertainty in forcing can be quantified by assessing changes in simulated streamflow on perturbing the forcing. Modeling hydrologic processes in a river basin can be a complex exercise because different hydrologic models incorporate different types of parameterizations to model the underlying processes. In this study, the footprint of uncertainty in observational datasets (e.g., IMD, TRMM, APHRODITE, GLDAS) on streamflow simulation is quantified using a suite of hydrological models (SWAT, HEC-HMS, EF5, WRF-Hydro) to obtain intra-model and inter-model daily streamflow ensembles for the Brahmaputra river basin. The response of the models to perturbed forcing and sensitivity of the hydrologic parameters to the simulated daily streamflow is explored. This approach helps in quantifying the observational and representational uncertainty of forcing, as well as the physical processes in the model. The model ensemble appears to capture the daily and intra-seasonal (flood-pulses) variations in the streamflow and gives quantifiably reliable simulations of daily discharge.