Key points are not available for this paper at this time.
Shape of mathematical curves describing local weather statistics are systematically influenced by large-scale conditions and geographical factors, and we present results suggesting that it is possible to downscale this kind of information directly. We downscale cumulative probability functionsfor daily precipitation and intensity-duration-frequency (IDF) curves for estimating return values of intense sub-daily rainfall. Recent progress in reanalyses such as ERA5 has made it possible to use wet-day frequency and mean precipitation intensity as predictors to compute local variations in wet-day frequency and mean intensity respectively, and we use a simple and approximate formula from earlier studies with these two parameters to specify the shape these curves. Downscaling the shape of such curves may be referred to as downscaling climate if we regard local climate as the statistical description of various weather parameters. This approach is distinct from the more traditional approach downscaling weather, where one seeks to estimate particular local states for instance on a day-by-day basis. We also present work on downscaling the shapes of pdfs and IDFs for large multi-model ensembles for the application in climate change adaptation efforts. Our eanalysis is accompanied by an evaluation of both methodology and the global climate models' (GCMs) ability to reproduce observed large-scale climatic variability in terms of the salient spatio-temporal covariance structure. When it comes to providing future regional climate projections, we also emphasise that it is important to combine different strategies for downscaling, e.g. regional climate models (RCMs) and empirical-statistical downscaling (ESD) that are based on different assumptions, for getting robust future regional climate projections.
Benestad et al. (Fri,) studied this question.