An evaluation of single-site statistical downscaling techniques in terms of indices of climate extremes for the Midwest of Iran

Seven single-site statistical downscaling methods for daily temperature and precipitation, including four deterministic algorithms [analog model (ANM), quantile mapping with delta method extrapolation (QMD), cumulative distribution function transform (CDFt), and model-based recursive partitioning (MOB)] and three stochastic algorithms [generalized linear model (GLM), Conditional Density Estimation Network Creation and Evaluation (CaDENCE), and Statistical Downscaling Model–Decision Centric (SDSM–DC] are evaluated at nine stations located in the mountainous region of Iran’s Midwest. The methods are of widely varying complexity, with input requirements that range from single-point predictors of temperature and precipitation to multivariate synoptic-scale fields. The period 1981–2000 is used for model calibration and 2001–2010 for validation, with performance assessed in terms of 27 Climate Extremes Indices (CLIMDEX). The sensitivity of the methods to large-scale anomalies and their ability to replicate the observed data distribution in the validation period are separately tested for each index by Pearson correlation and Kolmogorov–Smirnov (KS) tests, respectively. Combined tests are used to assess overall model performances. MOB performed best, passing 14.5 % (49.6 %) of the combined (single) tests, respectively, followed by SDSM, CaDENCE, and GLM [14.5 % (46.5 %), 13.2 % (47.1 %), and 12.8 % (43.2 %), respectively], and then by QMD, CDFt, and ANM [7 % (45.7 %), 4.9 % (45.3 %), and 1.6 % (37.9 %), respectively]. Correlation tests were passed less frequently than KS tests. All methods downscaled temperature indices better than precipitation indices. Some indices, notably R20, R25, SDII, CWD, and TNx, were not successfully simulated by any of the methods. Model performance varied widely across the study region.