*3.1. Empirical Quantile Mapping with a 90-Day Moving Window*

The application of eQM bias correction with a 90-day moving window improved the climate classification. RCMs simulated appropriate climate zones in each region with the exception of HIRHAM (Figure 3). In the case of HIRHAM RCM, the climate zone simulation was improved in the Northern regions, e.g, in Scandinavia, the British Isles and Mid-Europe, but it still produced dryer climate zones in the Iberian Peninsula, the Mediterranean and Eastern Europe. In the other RCMs, the extension of climate zones differed from the observed ones mainly in the Iberia Peninsula, in the Alps, in the Mediterranean and in Eastern Europe. The difference in the frequency of the occurrence of the climate zones was only 1–2% between the RCMs with exception of HIRHAM in each region. For example, the occurrence of the Cfb zone in the Alps was 53%, 51%, 50% and 50% according to the ALADIN, RegCM, RACMO2 and RCA simulations, respectively.

**Figure 3.** Simulated K-G climate classification according to E-OBS (**A**) and empirical quantile mapping (eQM) corrected precipitation and temperature with 90-day moving window in ALADIN (**B**), HIRHAM (**C**), RegCM (**D**), RAMCMO2 (**E**) and RCA (**F**).

The precipitation was mainly underestimated in each season except in the British Isles in winter (DJF) (Table 3a). The eQM with a 90-day moving window decreased the residual temperature bias in DJF but increased it in summer (JJA) in ALADIN, HIRHAM and RegCM in some regions (Table 3b).

**Table 3.** Residual bias of seasonal amount of simulated precipitation (a) and of seasonal mean of simulated temperature (b) in the case of eQM bias correction in eight different regions: the Alps (AL), the British Isles (BI), Eastern Europe (EA), France (FR), the Iberia Peninsula (IP), the Mediterranean (MD), Mid-Europe (ME) and Scandinavia (SC) in DJF and JJA The bias values are in % and in ◦C in the case of precipitation and temperature, respectively.



**Table 3.** *Cont.*

#### *3.2. Quantile Mapping Based on a Gamma + Generalized Pareto Distribution with a 90-Day Moving Window*

Owing to the fact the seasonal temperature probability distribution does not fit a Gaussian distribution due to non-Gaussian tails occurrence, the gpQM bias correction with a 90-day moving window was implemented only on the precipitation data. The gpQM with a 90-day moving window bias-corrected precipitation was combined with eQM with 90-day moving window corrected temperature values for the calculation of K-G zones. The gpQM correction with a moving window also improved the climate classification, but it resulted in dryer climate zones in some regions compared to the eQM correction (Figure 4). The Csb and BSk ratio was larger in the Mediterranean and Eastern Europe, respectively, according to the gpQM in each RCM. Owing to gpQM method the extension of dry zones (Csa, Csb, Dsb), the Csb zone was predominant in France, Mid-Europe and the Mediterranean, and the BSk was overestimated in the Iberian Peninsula and Eastern Europe in HIRHAM model.

**Figure 4.** Simulated K-G climate classification according to E-OBS (**A**) and quantile mapping of precipitation and temperature based on a gamma + Generalized Pareto Distribution (gpQM) correction of precipitation and eQM correction of temperature with 90-day moving window in ALADIN (**B**), HIRHAM (**C**), RegCM (**D**), RAMCMO2 (**E**) and RCA (**F**).

The residual precipitation bias was variable. The precipitation was overestimated in some regions, mainly in DJF. Although eQM correction resulted in better K-G classification, the residual bias of gpQM correction was smaller in some regions (e.g., in the Mediterranean in the case of RegCM) (Table 4).

**Table 4.** Residual bias in the seasonal amount of simulated precipitation (a) in the case of gpQM bias correction in eight different regions: the Alps (AL), the British Isles (BI), Eastern Europe (EA), France (FR), the Iberian Peninsula (IP), the Mediterranean (MD), Mid-Europe (ME) and Scandinavia (SC) in DJF and JJA. The bias values are in %.

