EVALUATING THE PERFORMANCE OF EFAS HYDROLOGICAL PREDICTIONS IN LATVIAN RIVER BASINS: A COMPARISON WITH OBSERVATIONAL DATA

This study evaluates the performance of the European Flood Awareness System (EFAS) [1] in predicting hydrological variables by comparing EFAS reforecast data with observational data from the Latvian Environment, Geology and Meteorology Centre (LVGMC). Using the open-source LISFLOOD hydrological model [2], the study examines the accuracy of ECMWF-driven predictions of river discharge and water levels across Latvia’s diverse river basins. The study employs a variety of interpolation techniques, including linear interpolation and nearest neighbour interpolation, to extract grid data from the Copernicus Early Warning Data Store (EWDS) [3] dataset at hydrological station points. To assess prediction accuracy, a range of statistical and error metrics, including Mean Error (ME) [4], [5], Root Mean Squared Error (RMSE) [5] - [7], Nash-Sutcliffe Efficiency (NSE) [5], [8]-[12] and Kling-Gupta Efficiency (KGE) [5], [12], [13], are utilized. The analysis highlights the effectiveness of EFAS in different seasonal and hydrometeorological conditions, identifying both strengths and limitations in the model's performance. Furthermore, the study explores potential calibration approaches to including regional forecasting capabilities, particularly in light of climate change impacts on low-flow and drought period predictions. This research provides valuable insights into the application of continental-scale hydrological models at the regional level, offering recommendations for improving the accuracy of flood forecasting systems.