Institute of Atmospheric Physics (IAP) AS CR - Czech Republic
Forecasting of local heavy convective precipitation, which can cause flash flooding, has been an important topic of investigation for a few decades. It has been recognized that non-hydrostatic models with resolution of the order of 1km (and less) are able to simulate the dynamics of organized convective systems provided that adequate triggering is entered in the model. Apart from experimental model runs there are also diagnostic case studies of convective events that employ such models and compare the results with e.g. radar data. Nevertheless, the purely deterministic prediction of severe convective systems and the corresponding quantitative precipitation forecast is far from being solved.
A national research project "Severe convective events in the region of the CR" has been running since 2000 with participation of the Czech Hydrometeorological Institute (CHMI) and IAP ASCR. In the block, devoted to the numerical modeling, convective events with heavy precipitation are studied. A fine resolution numerical weather prediction (NWP) model has been employed to assess its ability to simulate severe convective events as well as to evaluate the profit of the model application in routine precipitation forecasting and/or warning. In this contribution the results of model application will be summarized and discussed.
The non-hydrostatic NWP model LM DWD is employed, which was provided by German Weather Service (DWD) for research. The diagnostic run of the model consists of two steps. In the first step the driving LM model (LLM) is integrated over a large part of the Europe (2240×2016 km) with horizontal resolution 14km (161×145 grids), time step 150 s, and 20 vertical model levels. The initial and boundary conditions are derived from objective analysis of aerological data. The model is run with the convective parameterization of Tiedke. The nested LM model (SLM) with resolution 2.8 km, 137×137 horizontal grid points, and time step 15s is integrated over the area of interest. The SLM has 35 vertical model levels with the top at 16km above the sea level. The initial and boundary conditions are obtained from the LLM forecasts.
Historical convective event from July 1998, which caused an extreme precipitation amount and a flash flood over the NE part of the CR, was analyzed with several model runs. They differ in the design of the experiment (starting time of SLM integration, convective parameterization ON/OFF, modifications of model orography etc.). The LLM model forecasts yield reasonable larger scale meteorological fields including precipitation. The results of the SLM integration indicate that the inclusion of the convective parameterization smoothes precipitation fields, which is not realistic. The structure of precipitation fields obtained by the SLM with convective parameterization switched OFF better corresponds to the spatial structure of gauge and radar data. The time course of precipitation values depends on the start of the SLM model integration. The preliminary results show that the accumulated precipitation does not differ too much provided the SLM model starts its integration at least 6 hours before the event studied. The fact that the SLM model produces large precipitation in contrast to the routine forecast and that the area precipitation pattern corresponds to the radar precipitation is promising.
Similar model applications have been run for several convective events from 2000 and 2001. The set of events comprises cases documented in the project, where organized convective systems produced significant precipitation (heavy precipitation and/or hail). At present there are results from three 2000 events available, next integration including the events from 2001 is in process. The model results are compared with outputs of CHMI weather radars.
In the contribution the summary of the results will be presented and discussed.