Sources of Precipitation in Alpine Watersheds

Precipitation in a specified target area originates from various sources of moisture. These sources have to be properly described in numerical precipitation forecasting. In the context of a limited area model (LAM) these sources are initial moisture content and evapotranspiration within the model domain, and lateral moisture inflow into the model domain. It is the purpose of the proposed project to quantify the sensitivities of precipitation forecasts with regard to these quantities through numerical experiments with the LAM ALADIN (operational in Austria). The experiments shall reveal the relative importance of the individual sources of precipitation in a specified target area. At the same time the influence of remote areas on local precipitation will be quantified. This yields sensitivity patterns in space and forecast time. Several thousand model runs will be necessary in order to sample these patterns properly. This approach is reasonable with present day quality of LAMs and available computer resources. The target area of interest will be Eastern Austria; emphasis will be given to relevant flood events (August 2002, July 1997) and to events with similar synoptic weather conditions but without heavy rain and subsequent flooding. Evapotranspiration itself is sensitive to soil moisture, surface radiation and other meteorological parameters. The relevance of precipitation sensitivity to evapotranspiration and thus soil moisture is accepted at seasonal and climatological time scales. It is the working hypothesis of this proposal that the sensitivity of precipitation to evapotranspiration is not small compared to the other precipitation sources even on time scales of a few days. Additionally, in Alpine regions the uncertainties in soil moisture analyses and evapotranspiration forecasts are large. Therefore, special emphasis will be given to quantification of precipitation sensitivity to moisture sources in Alpine regions. If significant sensitivities to evapotranspiration in the Alpine area are found, then future developments in precipitation forecasting have to focus on land surface-atmosphere coupling in complex orography and thus on integrated hydrometeorological assimilation/forecasting systems improving soil moisture and evapotranspiration forecasts in Alpine watersheds. Therefore, the planned project will provide important guidance in future hydrometeorological modeling in order to improve precipitation forecasts. This project complements ongoing initiatives like ELDAS on the European scale for improving analyses of surface parameters. International collaboration is planned for both scientific and technical questions as well as the exchange of ideas and results. Ideally, this cooperation will also include the work on joint publications. To carry out the proposed research, funds at a total amount of EUR 127.236,- (for one post-doc, two students, facilities, material, and travel) are requested for the project duration of 2 years.

Modern weather forecasting increasingly utilizes forecasts from numerical prediction models. Their reliability rises on the one hand by continuous advancement of the models and on the other hand by additional and/or new observational data (e.g. from advanced satellites). Besides improving the forecast quality of large-scale features also the forecasting of local events more relevant for the public (e.g. events of heavy precipitation) is now possible. For this purpose, limited-area models (LAM) with high spatial resolution (1-10km grid spacing) are applied over a certain area (e.g. the Alpine region) and driven by data from a global model. The resulting high-resolution forecast is thus a complicated "mixture " of information from the global model and the LAM. Precipitation, albeit of high importance for the general public, is perhaps the most challenging quantity from a modeler`s point of view. This being a consequence of complicated (feedback) mechanisms in the atmosphere, which are represented in a simplified manner only within the model or of insufficient observational data (e.g. soil moisture measurements). The resulting numerical forecasts are thus associated with inherent uncertainties varying both in time and space. The objective assessment of the quality of numerical forecasts, denoted as evaluation in the following, is therefore a necessary and often very complex step during both the model development phase and also in an operational application. This project attempts to contribute to the improvement of numerical forecasts in at least to aspects: Firstly, the influence and importance of different sources of moisture is investigated. This contributes to a better understanding of the underlying processes and might also show potential deficiencies of the model. In a LAM the sources of moisture are the atmospheric moisture content at the beginning of forecast, the evapotranspiration within the model`s domain and the lateral moisture inflow into the model domain. The influence of these sources is examined through extensive numerical experiments. However the results indicate that the estimations are associated with large uncertainties and must be interpreted with great care. The second point concerns the evaluation of predicted precipitation fields. To this aim observations are necessary. Those are usually available from a limited number of gauges distributed unevenly in the low-lands. A quantitative comparison of spatial model forecasts and point measurements (i.e. from rain-gauges) is possible only if either the model forecasts are post-processed in a suitable manner ("downscaling to measurement sites"), or if the available observations are interpolated in space resulting in a precipitation analysis. Within this project existing methods have been developed further for application to high- resolution forecasts and have also successfully used for evaluation purposes.