Statistical Downscaling Techniques and Projection of Future Climate Extremes in the Texas Environment

This study evaluates the performance of statistical downscaling techniques using different metrics and compares extreme precipitation and temperature changes under climate change scenarios in the Bosque watershed, North-Central Texas, USA. The observed gridded Daymet data was used to downscale Global Climate Model simulations (GCMs) and evaluate statistical downscaling techniques. The mean, the 90th percentile, and wet day probability of climate model simulations downscaling were compared (before and after) with the Daymet data during the historical period (1981–2005). The frequency and intensity of extreme precipitation and temperature values were also analyzed under future climate scenarios (2031–2060 and 2070–2099). The GCMs (CCSM4, MIROC5, and MPI-ESM-LR) underestimate historical period mean monthly and annual precipitation. The Ratio Delta (DeltaSD) method effectively adjusts the mean monthly and annual precipitation, the wet day probability and the 90th percentile during the historical period. In contrast, the Quantile Mapping Method (QDM) is less effective to reproduce the mean and distribution-based metrics in the historical period. The changes in extreme values in the future climate were found to be highly influenced by the downscaling techniques than the driving GCMs and emission scenarios. The climate scenarios from DeltaSD reveal a low frequency of R10mm and R20mm days. In contrast, the climate scenarios developed from the QDM project have a higher frequency of R10mm, R20mm, and CDD. A consistent increase of maximum and minimum temperature and extreme temperature indices is projected in most climate change scenarios. The trend of extreme temperature indices shows sensitivity to emission scenarios where a significant (≤ 0.05) change in temperature extremes was detected under the RCP8.5 emission scenario.