Optimization of GIS-based Meteorological Forecast Systems

The usage of high-resolution digital geographic information bears a high potential for the improvement of the output of meteorological mesoscale forecast models which has in the past been exploited only to an insufficient extent. To face this problem, the combined meteorological and geographic information system MetGIS has been created within the framework of interdisciplinary, international research projects. MetGIS is fed by the gridpoint output of external forecast models such as GFS (Global Forecast System). These input data are downscaled to the very high horizontal resolution (100 m) of system-internal topographic data bases. Subsequently they are visualized and further manipulated through the core of the system, a sophisticated Graphical User Interface (GUI). Run in an automated mode, this GUI may also regularly generate a set of high-resolution, GIS-based forecast charts accessible via a web interface. The GUI includes a wealth of display options, styles and zoom facilities. The main topic of the proposed research project is the upgrading of MetGIS with innovative forecast and visualization technologies so that the system can be interesting also for future commercial application. Development work will focus on meteorological aspects (improvement of prediction quality) as well as on selected issues of data visualization. Among the principal goals of the project is the creation of detailed structures ("fingerprints") which will be derived form the available high resolution topographic data. This allows to improve the forecast of the snow line in a way, that the intra-alpine descent of this parameter in the case of heavy precipitation events can be predicted. The forecast of precipitation will be refined by the determination of flow-dependent climatological fingerprints. Concerning temperature, an improved detection of inversions by help of the high resolution topography is aspired. MetGIS will also be adjusted in a way that it can work with ensemble forecasts and involve these in the data downscaling procedures. The quality of the developed downscaling techniques will be validated by means of observation data bases constructed during the D-PHASE project, partly enhanced in quality by VERA (Vienna Enhanced Resolution Analysis). In the field of data visualization, the possibility to retrieve meteorological information related to individual points and polygons (e.g. highways) will be established.

High-resolution digital geographic information bears a high potential for the improvement of the output of numerical weather prediction (NWP) models. The described research project tried to exploit some of this potential, focusing on the improvement of MetGIS, a combined meteorological and geographic information system and downscaling (forecast refinement) approach that is based upon NWP model output. Development work concentrated on meteorological aspects (improvement of prediction quality) as well as on selected innovative issues of data visualization. It was processed in collaboration with project partners from Chile, Spain, Russia, India and Bhutan. These partners also provided sets of meteorological observation data which could used for verification and fine-tuning of the findings of the project. The project work started with the derivation of adequate structures (fingerprints) from the available highly-detailed topographic databases (horizontal resolution 30m). This allowed including also the shape of valleys in the forecast, beside altitude. Prediction of the snow line was improved: now the observed intra-alpine descent of this parameter in the case of heavy precipitation events can be simulated numerically. Besides, a new approach was developed to substantially improve the prediction of the temperature in valley and lowland locations, especially in the case of temperature inversions, using the simulated 2m-temperature of the driving NWP. Basic algorithms have been developed to include in high resolution downscaling systems the estimation of forecast probabilities. The procedures used were based on the Ensemble Prediction System (EPS) of the US National Weather Service and put a focus on the parameters air temperature and precipitation. In order to test to which degree the forecast quality of MetGIS depends on the driving model, a flexible interface for model input data was created, allowing MetGIS now to be operated not only on GFS forecasts, but also on the output of NWPs like ECMWF, ETA and WRF. In the field of the visualization of meteorological forecast data, a couple of innovative features were implemented. In web-based MetGIS forecast charts, the meteorological values related to arbitrary map locations can be displayed beside the mouse cursor. Point forecasts immediately pop up after clicking at any city, village or mountain name, and instant line forecasts (weather forecasts along routes) are constructed after selecting any mountain passes or mountain ascent paths. The output of the described research project has a high potential of improving operational weather forecasting systems. It may also have a positive impact on the performance of climate simulation models and produce a better meteorological input database for models from related fields (hydrological models, snow cover simulation models, etc.).