Stochastic rainfall generation accounting for climate changes

Funded by: FWO
Researcher: Willem Jan Vanhaute
Promotors: Niko Verhoest; Patrick Willems (KULeuven)
Begin date: 01/08/2011
End date: 31/07/2015

Rainfall is an essential variable of the hydrological cycle, which in turn is an indispensable part of many of the environmental sciences. Therefore, it is to everyone's best interest to have decent rainfall records at one's disposal. In some cases, the mere use of historical rainfall records can be justified. For certain other applications (e.g. hydrological design studies), a more extensive dataset will be required to obtain satisfactory results. In such cases, stochastic rainfall models can be regarded as valuable alternatives to using observed rainfall time series. An important prerequisite, however, is that these models have to be capable of simulating rainfall data exhibiting the same statistical properties as the observations at the location of interest.

Furthermore, increasing evidence suggests our climate is subject to changes affecting governing hydrological regimes. Such scenarios can be predicted using GCMs and LCMs. However, for practical applications the scale of these models is too coarse. Therefore, the aforementioned climate change scenarios can be introduced into the stochastic rainfall models to be able to assess the practical impact on a finer scale.

Specific research topics that will be addressed are:

-Comparison of state-of-the-art stochastic rainfall models (Bartlett-Lewis type)

-Improved calibration using global optimization techniques

-Improvement of model structure to enhance extreme value behaviour

-Incorporation of climate change scenarios into simulations