Observational trends, physical reasons and modelling results suggest an increase in extreme precipitation with climate warming. In particular, sub-daily precipitation extremes are expected to increase heavily raising concerns about the future impacts of flash floods in urban environments and for small or steep river catchments. In order to quantify the potential risk of flash floods in the future, impact studies often require site-specific sub-daily estimates of precipitation extremes. But in their current stage, most Regional Climate Models (RCMs) are only able to provide areal averaged projections at ca. 12.5km resolution and simulated sub-daily precipitation extremes tend to be heavily biased. As a result, statistical downscaling methods are needed to provide site-specific more reliable projections of sub-daily precipitation extremes. In this thesis, a statistical downscaling method was developed to project site-specific future hourly precipitation extremes over the UK. Circulation patterns (CPs) were classified using a fuzzy rules based approach to categorize extreme hourly precipitation events according to their corresponding atmospheric conditions. In a next step, an analogue day method was applied to find the most similar day in the past by comparing the RCM simulated daily precipitation and temperature with the observations for each CP. The daily maximum hourly precipitation record on the most similar day was extracted and perturbed based on precipitation duration-temperature relationships conditioned on CPs. Within the field of statistical downscaling techniques, the applied method is best described as a hybrid of the analogue and the regression-based method. It was shown that the method is capable of reproducing observed extreme hourly precipitation over different validation periods. Projections based on the applied statistical downscaling method indicate increases in UK hourly extremes but with high variations depending on the twelve different stations, the two future time periods, the two emission scenarios and the four different GCM-driven RCMs.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:699582 |
| Date | January 2016 |
| Creators | Rau, Markus |
| Publisher | University of East Anglia |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | https://ueaeprints.uea.ac.uk/61048/ |
Page generated in 0.0016 seconds