OPAQUE -

Operationelle Abfluss- und Hochwasservorhersage in Quellgebieten   operational discharge and flooding predictions in head catchments

Overview

Working package 2

Operational estimation and short-term prediction of the area precipitation

The second package aims at an operational short time prediction of precipitation in the catchment. The estimation is based on two sources of information: Rainfall radar and ground station measurements. Predicted rainfall is used for the operational simulation of the discharge in working package 3. In case of warnings for heavy precipitation the hydrologic models will be driven by the operational short time prediction for rainfall from the so-called "on-line Merging". The procedure is based on a method developed by EHRET (2002) at University of Stuttgart which combines the advantages of both measuring systems, i.e. the accuracy of the ground station data and the spatial distribution of the radar measurements. In this project, the first step is a systematic analysis of the existing merging procedure. Analyses are carried out in the catchments of the upper Danube, the upper Iller and the Weißeritz for different precipitation types and weather conditions over an extended period. The following improvements are planed:

• weighting method considering data quality of radar data
• estimation of precipitation in the case of missing values
• improved parameters (e.g. form parameters of the arc tangent function) based on longer time series
• increased computing speed
• preferential use of correlated radar and station data
• estimation of the precipitation variance within the raster cells by extrapolation
• reduction of errors by spatiotemporal interpolation of precipitation fields.

Additional to the LM forecast a short-term prediction over 2-3h, a so-called Now-casting of the local precipitation distribution on basis of radar data will be developed. The impossibility to predict the precipitation situation in the radar image already for the duration of few hours accurately suggests a stochastic prediction approach. Thus one is able to count ensembles and to win reference points over the range of possible developments. The SCM procedure of Ehret (2002), based on combined radar and station measurements, will be enhanced in the project regarding the wind field as function of place and time, the hierarchical modeling on use of Germany-wide radar compositions, the parameterization of the transition matrices for the description of the temporal development of precipitation intensity on the scale of a radar pixel as a function of precipitation type (convective or stratiform) and regarding the incident flow and the consideration of orographic effects.