Short and Medium Range Hydrometeorological Ensemble Prediction System: A mesoscale case study in Denmark
Abstract
Recent floods in Europe have increased the need to improve flood prediction systems that will allow, firstly, for proper analysis of the uncertainties in the hydrological predictions and, secondly, for the generation of probabilistic forecasts. The use of Meteorological Ensemble Prediction Systems (MEPS) to propagate climate input uncertainty throughout the hydrological modeling process is a state-of-the-art methodology for issuing probabilistic forecasts. MEPS have been made operational at meteorological centers around the globe for both short and medium range forecasting. The European Centre for Medium-Range Weather Forecasts (ECMWF) issues probabilistic forecasts with a resolution of approximately 25 kilometers using 50 ensemble members. The forecasts are issued twice per day with a lead time up to 15 days. Likewise, the Danish Meteorological Institute (DMI) issues probabilistic forecasts for a limited area covering Northern Europe with 25 ensemble members. DMI's MEPS has a spatial resolution of 5 kilometers, a lead time of 48 hours and is issued four times per day. Our research focuses on the implementation of both MEPS as input variables in a hydrological forecast model. We are aiming at generating seamless probabilistic forecasts of hydrological variables such as streamflow and groundwater levels for Silkeborg city in Denmark. A MIKE SHE coupled surface-groundwater hydrological model set up for the Silkeborg catchment is used for the hydrological forecast. Short and medium range forecasts for streamflows and groundwater levels of past events are analyzed in order to study the feasibility and skills of the flood forecasting system. Several measures of skill are described and analyzed in order to detect biases and lack of accuracy in our forecast system. A study of the reliability, skill over climatology/persistence forecasting, resolution, discrimination, etc., was made by means of the analysis of the joint distribution of the forecasts and observations of hydrological variables. The innovative value of this study encompasses three main issues. The first one is the use of integrated surface/groundwater modelling. Most of the literature focuses on the study of the reliability of the forecasts for streamflow only. In this study forecasting of other hydrological variables, such as groundwater levels are explored. The second one is the small-scale of our case study. The Silkeborg model covers an area of only 144 km2. Additional challenges apply for this situation given that the available information is limited. The third one is the inclusion of the DMI short-range, limited area MEPS in the system. Additional fuzzy skill measurements are computed for both meteorological and hydrological variables to mitigate punishments when rain events are predicted in a vicinity area of the grid of the observed event.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2013
- Bibcode:
- 2013AGUFM.H12A..03L
- Keywords:
-
- 1816 HYDROLOGY Estimation and forecasting;
- 1840 HYDROLOGY Hydrometeorology;
- 1873 HYDROLOGY Uncertainty assessment