Evaluating the uncertainty of climate model structure and bias correction on the hydrological impact of projected climate change in a Mediterranean catchment

Study region: Crati River Basin, Southern Italy, Central Mediterranean. Study focus: We evaluate the combined effect of multiple global and regional climate model (GCM-RCM) combinations and bias correction (BC) methods on the hydrological impact of projected climate change. Under the representative concentration pathway RCP4.5, 15 EUROCORDEX members, combining 6 GCMs and five high-resolution (0.11?) RCMs, provide the meteorological input for a spatially distributed hydrological model. RCM-derived input data are uncorrected and corrected through three empirical methods, leading to 60 different simulations for three ~30-year future periods in 2020-2096, compared to the baseline 1975-2005. The combined uncertainty of the climate models and correction methods is evaluated for the main hydrological variables using an analysis of variance (ANOVA) method. New hydrological insights for the region: Results highlight a considerable agreement in projecting a decreasing trend of available water resources (on average,-70 % for snow,-8 % for root zone soil moisture and -17 % for river runoff in the period 2070-2096), due to the remarkable mean temperature increase and less accentuated precipitation reduction. The uncertainty evaluation shows that (1) the primary source of uncertainty is the driving GCM, and (2) BC methods smooth the projected hydrological impact in a not negligible way, especially concerning discharge (for each future period, the reduction projected without bias correction is about 3 % higher than with BC), therefore contributing to the total uncertainty.

Automated summary

By evaluating the combined effect of multiple climate models and bias correction methods, the study found that the water resources in the Crati River Basin in Southern Italy will decrease significantly in the future decades, mainly due to the increase in temperature and the decrease in precipitation. The driving climate model was identified as the primary source of uncertainty, and the bias correction methods contributed to reducing the uncertainty.