Automatic Parameter Identification for a Rainfall-runoff Model within the European Flood Alert System

In this paper we address the problem of parameter identification and parameter uncertainty estimation for the rainfall-runoff model LISFLOOD. This model forms the core of the European Flood Alert System (EFAS), an activity that forecasts floods at the European scale. LISFLOOD is driven by meteorological input data and simulates river discharge in large drainage basins as a function of spatial information on topography, soils and land cover. Even though LISFLOOD is physically-based to a certain extent, some processes are only represented in a lumped conceptual way. As a result, some parameters lack physical basis and cannot be directly inferred from quantities that can be measured. In the current LISFLOOD version five parameters need to be determined by calibration. We outline an automatic calibration procedure for the LISFLOOD model used within EFAS. The Shuffled Complex Evolution Metropolis (SCEM-UA) global optimization algorithm is used to automatically search through the space of feasible parameter values and to obtain the combination of parameter values that produces the best model performance. Within one optimization run SCEM-UA also yields the underlying posterior parameter distribution, which reflects the uncertainty about the unknown model parameters. As an illustrative example, we demonstrate the methodology for the Meuse catchment upstream of Chooz, covering an area of approximately 8,850 km.