Downscaling heavy precipitation over the United Kingdom: a comparison of dynamical and statistical methods and their future scenarios

Abstract

Six statistical and two dynamical downscaling models were compared with regard to their ability to downscale seven seasonal indices of heavy precipitation for two station networks in northwest and southeast England. The skill among the eight downscaling models was high for those indices and seasons that had greater spatial coherence. Generally, winter showed the highest downscaling skill and summer the lowest. The rainfall indices that were indicative of rainfall occurrence were better modelled than those indicative of intensity. Models based on non-linear artificial neural networks were found to be the best at modelling the inter-annual variability of the indices; however, their strong negative biases implied a tendency to underestimate extremes. A novel approach used in one of the neural network models to output the rainfall probability and the gamma distribution scale and shape parameters for each day meant that resampling methods could be used to circumvent the underestimation of extremes. Six of the models were applied to the Hadley Centre global circulation model HadAM3P forced by emissions according to two SRES scenarios. This revealed that the inter-model differences between the future changes in the downscaled precipitation indices were at least as large as the differences between the emission scenarios for a single model. This implies caution when interpreting the output from a single model or a single type of model (e.g. regional climate models) and the advantage of including as many different types of downscaling models, global models and emission scenarios as possible when developing climate-change projections at the local scale.

Linking climate change modelling to impacts studies: recent advances in downscaling techniques for hydrological modelling

“…Similar conclusions were derived by comparing dynamically and statistically downscaled precipitation and temperature time series for three mountainous basins in Washington, Colorado and Nevada, USA (Hay and Clark, 2003). Haylock et al (2006) compared six statistical and two dynamical downscaling methods with regard to their ability to downscale seven indices of heavy precipitation for two station networks in northwest and southeast England. Generally, winter showed the highest downscaling skill and summer the lowest; skill increases as the spatial coherence of rainfall increases.…”

“…However, traditional methods, such as stepwise regression, compositing, correlation analysis, PCA and CCA were more useful than novel methods such as ANNs (e.g. Harpham and Wilby, 2005); although Haylock et al (2006) developed a novel approach within one of the ANN methods to output the rainfall probability and gamma distribution scale and shape parameters for each day which allowed resampling methods to be used to improve the estimation of extremes. Station to station differences were found to dominate index to index, method to method and season to season differences, although methods generally performed better in winter than summer, particularly for precipitation.…”

“…Watts et al (2004) presented similar conclusions for the CRU WG for extremes based on 90th and 95th percentile values. Downscaling methods were also shown to capture some processes better than others (Haylock et al, 2006;Goodess et al, 2007); indices representing the frequency of extremes were better reproduced than those for the magnitude of events. Occurrence processes are better captured by the downscaling process as predictors based on large-scale circulation capture the patterns prohibiting precipitation occurrence whereas event magnitude depends on smaller local-scale processes.…”

Simulation of seasonal precipitation and raindays over Greece: a statistical downscaling technique based on artificial neural networks (ANNs)

“…This choice was made for compatibility with the perfect boundary conditions . It should also be noted that the aforementioned periods (calibration-validation) were selected in the context of the STARDEX project and were employed in all the publications derived from that project (http://www.cru.uea.ac.uk/projects/stardex, Goodess et al, 2006;Haylock et al, 2006). Although the use of 15 years as a validation period is quite short, given the available time series it was not possible to use a longer validation period.…”

Circulation types classification for hourly precipitation events in Lublin (East Poland)

“…Heavy rains increase the threat of floods, often causing deaths and substantial financial losses. Such events in recent years have led to an increase in the number of studies concerning changes in extreme precipitation events in various regions of the world [1][2][3][4][5][6].…”