Abstract
Both statistical and dynamical downscaling methods are well established techniques to bridge the gap between the coarse information produced by global circulation models and the regional-to-local scales required by the climate change Impacts, Adaptation, and Vulnerability (IAV) communities. A number of studies have analyzed the relative merits of each technique by inter-comparing their performance in reproducing the observed climate, as given by a number of climatic indices (e.g. mean values, percentiles, spells). However, in this paper we stress that fair comparisons should be based on indices that are not affected by the calibration towards the observed climate used for some of the methods. We focus on precipitation (over continental Spain) and consider the output of eight Regional Climate Models (RCMs) from the EURO-CORDEX initiative at 0.44∘ resolution and five Statistical Downscaling Methods (SDMs) —analog resampling, weather typing and generalized linear models— trained using the Spain044 observational gridded dataset on exactly the same RCM grid. The performance of these models is inter-compared in terms of several standard indices —mean precipitation, 90th percentile on wet days, maximum precipitation amount and maximum number of consecutive dry days— taking into account the parameters involved in the SDM training phase. It is shown, that not only the directly affected indices should be carefully analyzed, but also those indirectly influenced (e.g. percentile-based indices for precipitation) which are more difficult to identify. We also analyze how simple transformations (e.g. linear scaling) could be applied to the outputs of the uncalibrated methods in order to put SDMs and RCMs on equal footing, and thus perform a fairer comparison.
Highlights
Different climate downscaling techniques have been developed since the early 1990s to bridge the gap between the large-scale climate information provided by GlobalCirculation Models (GCMs) and the regional-to-local scale required for climate impacts assessment
It is important to note that the EURO-CORDEX Regional Climate Models (RCMs) have not assimilated any information from Spain044 observations, whereas the Statistical Downscaling Methods (SDMs) have been cross-calibrated using them —in particular, Generalized Linear Models (GLMs) are trained minimizing the distance between the observed and predicted/downscaled daily mean training error.— RCMs typically exhibit non-negligible biases (Casanueva et al, 2015), whereas mean precipitation is usually well represented by the different SDMs
In order to give a spatially averaged measure of accuracy avoiding the compensation of opposite sign biases, we use throughout the entire paper the spatially averaged mean absolute error (MAE), which is calculated as the spatial average of the absolute value of the mean temporal errors at each grid box
Summary
Circulation Models (GCMs) and the regional-to-local scale required for climate impacts assessment (see Maraun et al, 2010, and references therein). Unlike RCMs, SDMs are calibrated in a training phase using some sort of optimization or resampling process (or establishing a correction function in bias correction methods) involving the available observations (see e.g. Maraun et al, 2010) As a result, these methods are trained with local observations to reproduce some observed statistics, which are directly affected by the particular calibration process (i.e. optimization, re-sampling, or distribution-mapping process). SDMs provide information at the spatial scale given by the observations (i.e. point stations or grids), whereas RCM results are areal-representative (of the model grid boxes) and, cannot represent the local variability of point stations (Luo et al, 2013) For this reason, recent studies acknowledge that a fair comparison of RCMs and SDMs requires the use of observational gridded data sets for SDMs calibration and both techniques evaluation (Schmidli et al, 2007; Hertig et al, 2012; Ayar et al, 2015).
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
