*3.3. Uncertainty in Freshwater Changes*

The error bars in Figures 6 and 7 show the uncertainty ranges of the estimated freshwater changes Δ*ET* and Δ*R*, as obtained in total from all considered uncertainty scenarios (Table 3, Supplementary Table S1). Table 4 lists further result details for the total regional catchment; corresponding results for the other catchments are listed in Supplementary Table S2 for Mainland, Table S3 for Peloponnese, Table S4 for Ionian and Table S5 for Aegean. In general, the total uncertainty range of Δ*ET* (±6–8 mm/year across all catchments) is smaller than that of Δ*R*, which is also more variable among the catchments (from ±8 for the Aegean up to ±18 mm/year for the Peloponnese). Overall, these uncertainty ranges do not change the main results and implications discussed above based on the mean freshwater changes.

With regard to the investigated underlying uncertainties, the largest range is associated with the different scenarios of irrigation water use per irrigated area (*Iww*; Table 3). For the total regional catchment (Table 3), the estimated change in *Iww* is on average 416 mm/year and the associated uncertainty range is ±156 mm/year. In comparison, the resulting uncertainty ranges for this catchment are much smaller for both Δ*ET* and Δ*R*, at ±7 and ±10 mm/year, respectively. Similar results are obtained for this uncertainty propagation in all catchments, implying a major decrease in absolute range magnitude from that of underlying uncertainties (Table 3, Supplementary Table S1) to that of the total resulting Δ*ET* and Δ*R* uncertainties (Table 4, Supplementary Tables S2–S5).

Also in relative terms, the largest underlying uncertainty range is that for the *Iww* change at ±38%. For the decrease in *R* (Figure 7), the resulting relative uncertainty range is overall smaller: 13% for the total regional, Mainland and Aegean catchments (with mean Δ*R* of −75 mm/year, −73 mm/year and −65 mm/year, respectively); 15% for the Ionian catchment (with mean Δ*R* of −119 mm/year); and 17% for the Peloponnese catchment (with mean Δ*R* of −91 mm/year). For the increase in *ET* (Figure 6), the relative uncertainty range is larger and varies more among the catchments than that of Δ*R*. Specifically, the relative Δ*ET* range is: ±28% for the total regional, ±25% for the Mainland and ±23% for the Aegean catchment (with mean Δ*ET* of 26 mm/year, 29 mm/year and 28 mm/year, respectively); ±54% for the Ionian catchment (with mean Δ*ET* of 16 mm/year); and ±194% for the Peloponnese catchment (with mean Δ*ET* of just 4 mm/year).

In general, the largest contribution to the total Δ*ET* range (of ±6–8 mm/year) stems from irrigation uncertainty (±5–6 mm/year contribution; with additional ±1 mm/year stemming from climate

uncertainty). For the total Δ*R* range (of ±8–18 mm/year), the largest uncertainty contribution varies among catchments. For the Aegean, Mainland and total regional catchment, the largest contribution stems from irrigation uncertainty (±6–8 mm/year contribution to their total Δ*R* range of ±8–10 mm/year; with additional ±2–4 mm/year stemming from climate uncertainty, in which the main part is due to the Δ*P* uncertainty of around ±2 mm/year). In contrast, for the Ionian and Peloponnese catchments, the largest contribution stems from climate uncertainty (±9–11 mm/year, in which the main part is due to the Δ*P* uncertainty of ±7–8 mm/year, contributing to the total Δ*R* range of ±16–18 mm/year; with ±6–7 mm/year stemming from irrigation uncertainty).

Across all catchments, the scenario combination with maximum *Iww1* and minimum *Iww2* (i.e., Alt. 2 *Iww* from Table 3) yields the lower Δ*ET* and the upper Δ*R* limit, while the opposite scenario combination with minimum *Iww1* and maximum *Iww2* (Alt. 3 *Iww*, Table 3) yields the upper Δ*ET* and the lower Δ*R* limit. The actual values of these limits depend also on the considered Δ*P* scenario, with the undercatch correction (Alt 1 *P*, Table 3) and even more so the undercatch and orographic correction (Alt 2 *P*, Table 3) increasing the Δ*ET* values and decreasing the Δ*R* values compared to the base case scenario. The alternative scenarios for *ET*/*P* (Supplementary Table S6) and *αirr* (not shown) do not significantly change the results compared to the Base case scenario.
