Calibration and evaluation of the global flood forecasting system using daily streamflow observations

Global Flood Awareness System (GloFAS) produces ensemble streamflow forecasts and threshold exceedance probabilities for large rivers worldwide. The system employs a coupled land surface (H-TESSEL) and flow routing (LISFLOOD) models for simulating the hydrological processes at a gridded scale. The flow routing model has been using a priori parameter estimates with uniform values over a large spatial area, which can potentially lead to poor streamflow forecast skill. The objective of this work is to improve the GloFAS forecast through calibrating the global LISFLOOD model parameters. We evaluate the resulting improvements by computing the skill scores relative to streamflow simulations using default model parameters. We present results from the calibration and evaluation using daily streamflow observations from 1167 stations around the world. Future works include regionalization of model parameters in order to improve the simulation skills for basins without streamflow data, and calibration of the land surface model parameters.