Model-based prediction of area weather events

An important topic in meteorology is the computation of probabilities for the occurrence of certain weather events, e.g. (heavy) precipitation. There exists a variety of numerical and probabilistic methods to compute Point probabilities, i.e. probabilities that the considered weather event occurs at a given location. Additionally, meteorologists are interested in area probabilities, i.e. probabilities for the occurrence of considered weather events somewhere inside a given region. In literature, however, no solution seems to be known in order to derive area probabilities from point probabilities analytically or by stochastic simulation.

In a joint research project with the German Meteorological Service (Deutscher Wetterdienst - DWD), we therefore develop a model-based approach for the computation of area precipitation probabilities. We construct a spatial stochastic model for the representation of precipitation cells and, related to that, for the representation of (point and area) precipitation probabilities. The model parameters are estimated statistically from a set of point probabilities that are provided by DWD. Then, the specified model allows for the computation of area precipitation probabilities for arbitrary areas of interest either analytically or by repeated Monte Carlo simulation. Further planned research includes the integration of exceedance probabilities for certain precipitation amounts into the model.

Possible topics for theses include:

  • Improvement of the model for precipitation probabilities by developing and  implementing further model components

If you are interested, please contact Prof. Schmidt or Peter Schaumann.

Back to projects.

Back to theses.