NWP-model verification over complex terrain with VERA

Since 2003 an operational NWP-model comparison system for several well-known weather prediction models (ALADIN, ECMWF, LME) has been embedded in the analysis system VERA (Vienna Enhanced Resolution Analysis) at the Department for Meteorology and Geophysics, University of Vienna. In the course of the international project MAP D-PHASE (2007) the system was rebuild and enhanced for further models (COSMO, MM5, GEM-LAM) on a new version of VERA, VERAXX. It enables to flexibly define different spatial resolutions (up to 2km) and geographical domains and, thus, provides a tool for an improved investigation of scale-dependent phenomena. In the framework of VERITA several high-resolution NWP-models shall be evaluated using modern state-of-the- art verification methods. In particular, methods dealing with the scale-decomposition of model fields and the separate treatment of different scales as well as the definition and propagation (in model chains) of phase errors shall be developed, tested and implemented. Further, questions concerning the usefulness of VERA for verification purposes in comparison with, on the one hand, alternative analysis methods and, on the other hand, meteorological observations shall be faced and answered. It shall be determined if analyses and (irregularly distributed) station observations can be used equivalently and, if yes, from which spatial scale, which station density and model resolution. The data basis for the project (high-resolution forecast models and a dense observation network) is provided by two international programs to which, both, the IMGW is strongly contributing: In the framework of COPS (Convective and Orographically-induced Precipitation Study) a major measuring campaign was performed from June to August 2007 in and area enveloping the Vosges, the Rhine Valley, the Black Forest region and the Swabian Alb. The main aim of COPS is to gain an enhanced knowledge on convective atmospheric processes. This is essential in order to improve quantitative precipitation forecasts of current high- resolution models and to find out about model errors and weaknesses. At the same time as COPS the demonstration period of the Forecast Demonstration Project (FDP) MAP D-PHASE (Demonstration of Probabilistic Hydrological and Atmospheric Simulation of flood Events in the Alpine region) took place in the period of June to November 2007. In the course of the demonstration phase an end-to-end forecast chain leading from modellers to forecasters and finally to end users was operated for heavy precipitation events and flowing events. The model forecasts, which were run by different meteorological institutions during the D-PHASE period in the Alpine Region as well as the measuring results of COPS, are available for all project partners for further scientific applications. In order to enhance the observation density in the whole Alpine Region, additional observation data (non-gts) where required by all national weather services around the Alps.

In the project VERITA not only forecasted field from models have been verified but also the reference fields used for verification have been quality controlled. Hitherto the reference data (observations or analysed fields) have been regarded as the truth. But these data also contain uncertainties which influence the verification results. Up to now, these errors have been assumed to be small compared to the errors in the forecasts. With increasing forecast accuracy however, this assumption does not hold any more. In the course of the project methods have been developed to estimate the accuracy of the reference data by calculating an Analysis Ensemble. By using this ensemble uncertainty estimates of the verification measures are possible for the first time.In a second step a comprehensive forecast model comparison study has been performed. This has become possible since the project has been embedded in two projects of the world weather research programme (WWRP) from the World Meteorological Organisation (WMO) a specialised agency of the United Nations. With the support of the WWRP projects a very dense observation data set and forecast data from a model ensemble over Central Europe could be collected. A model independent analysis has been conducted by using the VERA system (Vienna Enhanced Resolution Analysis) originally developed at the Department of Meteorology and Geophysics at the University of Vienna. These data sets have been used to verify model chains from different countries and quantify their performance for the first time.A further focal point has been put on the development of new verification methods and on the use of other verification parameters than usual. Precipitation is by far the most used verification parameter. As new parameters have been chosen: equivalent potential temperature (a measure for air masses), winds, frontal location and frontal speed. To assess the additional information of the new generation of high resolution forecast models a wavelet analysis has been utilized. Thereby, a scale analysis is performed on the forecasted and analysed fields. This allows to break down the differences of the models and the analysis and the differences between the models, respectively, on single scales. A detailed knowledge of the model performance is crucial for the forecaster on duty as well as for the modeller to analyse the weaknesses of the model and to increase its accuracy. Accurate weather forecasts are important in many areas of a national economy.