CONFERENCEAnalysis of short-term hydrological forecasts from different numerical weather models
Andrejs Zubanics
Latvian Centre for Environment, Geology and Meteorology
Liga Klints
Latvian Centre for Environment, Geology and Meteorology
DOI: 10.22364/fg.15.16
Keywords: hydrological ensemble forecasts, numerical meteorological models, comparative analysis
Abstract
Short-term hydrological forecasts are one of the most useful tools in flood situations, providing information two days in advance. Weather forecasts provided by several NWP models are widely used data for hydrological forecasting. In this research, HIRLAM and ECMWF weather forecasts were used. The aim of the research was to evaluate which weather forecasts could provide better assessment of hydrological conditions. Hydrological forecasts were calculated for three Latvian river catchments: Abava – Renda, Gauja – Sigulda and Dubna – Sīļi, representing regional differences in the territory of Latvia. Time periods were chosen to reflect the flooding period of the year 2013 and autumn flash floods of the year 2014. The modelling process was carried out using the semi-distributed conceptual IHMS-HBV model. The model calibration results for respective river catchments were evaluated with the NashSutcliffe efficiency coefficient with values varying between 0.86 and 0.88. The acquired forecasts were compared to real observations and against each other. By conducting a comparative analysis, it was ascertained that slightly better results in precision and timing on volume of the water in all three studied river basins were obtained using ECMWF weather forecasts. Greater errors for all hydrological forecasts were on the second day of the forecast.