A limited-area short-range ensemble prediction approach for heavy rain events
Abstract
Quantitative precipitation forecasts (QPF) from numerical weather prediction (NWP) models exhibit a significant uncertainty, which is usually being addressed by an ensemble approach. The interest of doing this with a high-resolution limited-area NWP model (LAM) has grown during the last years, and there are many different possibilities of designing a LAM ensemble. The design may depend on the application of the ensemble predictions later on. While it is often attempted to address many aspects of severe weather with an ensemble, this work investigates the potential of an ensemble design targeted to rainfall events, that can lead to river flooding. The work is part of the EU funded project "An European Flood Forecasting System" (EFFS), where the LAM ensemble predictions are to be applied in hydrological models in order to assess water levels of European rivers. Two different approaches of creating a small LAM ensemble for rainfall prediction are investigated. They are both based on the Danish Meteorological Institute's (DMI) version of the HIgh Resolution Limited-Area Model DMI-HIRLAM, which receives initial as well as lateral boundary data from the ECMWF ensemble prediction system. The first ensemble is designed to represent an uncertainty in the initial condition (atmospheric state) as well as at the lateral LAM boundaries. It is created on basis of the ensemble prediction system (EPS) from ECMWF, from which a small set of significant members is selected. The significance of a member is estimated on basis of accumulated precipitation over the area of interest, which includes the river basin under consideration. The selected ECMWF-EPS members make up the initial and boundary data for the members of the first LAM ensemble. The second LAM ensemble consists of the variation of the physical representation of the condensation and convection processes within the LAM, where several parameterization schemes available in DMI-HIRLAM are utilized. The two ensemble designs are applied on three historical heavy rain events of different synoptic and dynamic characteristics. Verification against precipitation analyses from the Deutsche Wetterdienst (DWD) show the capabilities of the ensembles in forecasting the severe rain events.
- Publication:
-
EGS - AGU - EUG Joint Assembly
- Pub Date:
- April 2003
- Bibcode:
- 2003EAEJA.....1657S