LINK – Analysis and nowcasting of extreme weather events based on microwave link data
Prediction improvement of extreme weather events by deriving precipitation values from microwave link data by methods of artificial intelligence.
There is scientific consensus that the trend towards a significant rise of extreme weather events due to long-term climate changes continues. A localised and timely accurate short-term prediction for the next hours of such weather events is a great advantage for the population and the economy. It allows the implementation of appropriate countermeasures to minimize damage and a better planning and reaction in the case of weather related extreme situations. Short-term predictions require a dense network of measurements for the provision of up-to-date weather data. Such measurements may stem from either ground based stations or remote systems like weather radar or satellites. In large areas of Austria, however, this is only partly possible with sufficient coverage: the number of ground stations is limited, especially in rough terrain, and radar data may lack in certain regions due to topographical reasons.
The main research idea of the present project is the use of physical data from commercial microwave link (CML) networks as used in mobile telephony to gain precise local rainfall information. The continuous extension of CML networks generates area-wide and large amounts of data about the physical properties of radio transmissions. Network operating companies routinely measure these properties (such as the loss in signal strength during transmission between antennas) in order to guarantee network quality.
The goal of the project is the evaluation of the usability of CML data for the prediction of extreme weather events. To achieve this goal, the following scientific problems must be solved:
- The project must analyse raw data from CML with respect to its quality and its error characteristics. To this end, the project will use artificial intelligence methods, especially machine learning methods.
- The adjusted, pre-processed, and aggregated data must be assimilated into numerical weather prediction models integrated into prediction models
- The benefit of the resulting analysis and forecast data sets have to be evaluated and validated in cooperation with the public customer involved in this project.
Oliver Eigner / Institute of IT Security Research, UAS St. Pölten
Zentralanstalt für Meteorologie und Geodynamik – ZAMG
Hutchison Drei Austria GmbH
UAS St. Pölten GmbH, Institute of Media Economics
Amt der Steiermärkischen Landesregierung, Abteilung 14 – Wasserwirtschaft, Ressourcen und Nachhaltigkeit, Referat Hydrographie