"Machine Learning, Meteorology and … data!" Dr. Irene Schicker from ZAMG

Machine Learning, Meteorology and … data – Use cases and studies at the Zentralanstalt für Meteorologie und Geodynamik (ZAMG)

Meteorology and big data were and are closely related. With the deployment of meteorological observation networks in the 18th century, their growing site density and at least five measured parameters, the development of numerical weather prediction (NWP) and climate models and remote sensing instruments in the early 20th century the amount of available data increased. One could thus conclude that huge amounts of data are nothing unusual in meteorology. Adding now the number of available, “crowdsourced” measurements and measurements at renewable energy sites the available potential of usable data explodes.

NWP models make use of sub-surface, surface and upper air observations, remote sensing data and recently also of crowdsourced data. With spatial resolutions of up to 1 km NWP models are able to provide even more detailed information. However, to provide tailored forecasts even such high resolutions of the NWP models are often too coarse to provide tailored information required by e.g. the industry, road services, or renewable energy sites. Other methodologies are, thus, needed.

In recent years, machine learning based forecasting tools were and are continuously developed at ZAMG to provide operational forecasts for wind speed and gusts at meteorological sites and nacelle height of wind turbines. Ongoing research is carried out to implement more sophisticated neural networks for wind and solar energy as well as for precipitation and temperature short-range forecasting on a grid using at least 1 km resolution. Additionally, clustering algorithms are investigated for unsupervised detect of temperature clusters in Austria. An overview on current work and research at ZAMG will be given.

Irene Schicker studied Meteorology in Vienna and Innsbruck and received her Master’s degree in the field of remote sensing and glaciology. Her PhD at the University of Natural Resources and Life Sciences, Vienna, covered high resolution numerical weather prediction simulations and some model adaptations. In 2014 she left the University to work at the Zentralanstalt für Meteorologie und Geodynamik where she gradually moved away from NWP towards renewable energy and machine learning.