*2.5. Uncertainty Analysis*

For comparison with and uncertainty analysis of other possible variable values than those considered in the base case scenario (Table 2), we also evaluate alternative scenarios of uncertain variable values. Table 3 compares the base and the alternative scenarios and lists their variable value differences for the total regional catchment; Supplementary Table S1 does the same for the other four study catchments.

With regard to *P* data, the alternative scenarios outlined in Table 3 (named in the first and explained in the second column, in direct comparison with the base case) consider possible corrections for *P* undercatch bias [37], or for both undercatch and orographic biases [38]. Moreover, the *ET*/*P* ratio used for estimating average *ET* in the recent period 1990–2009 ranges from 0.55 to 0.72 among the local catchments with available data across Greece (Table 1). The base case scenario considers the mean value of *ET*/*P*, whereas two alternative *ET*/*P* scenarios consider corresponding minimum and maximum values (Table 3); the *ET*/*P* values are multiplied with *P* in each data grid cell to obtain grid-cell values of *ET* over each catchment.

Furthermore, considerable uncertainty about irrigation variables may stem from the estimation of water use (withdrawal) per irrigated area in the 1990–2009 period (*Iww2*). For example, for the total regional catchment (Table 3), *Iww2* may be as large as 747 mm/year if the recent irrigated area in its calculation is estimated from the national reported irrigated area [30] (*Aai2* = 13,963 km2) instead of 591 mm/year in the base case scenario that estimates *Iww2* from GMIAv5.0 [32] (*Aai2* = 11,046 km2). The resulting uncertainty range of *Iww2* is then ±78 mm/year.

With this uncertainty range given for *Iww2*, a corresponding uncertainty range of ±78 mm/year may also be expected for *Iww1*, yielding an even wider uncertainty range for the temporal change in *Iww* by considering different possible combinations of *Iww2* and *Iww1* values in the base and alternative scenarios (as outlined in Table 3). Finally, also the value of the irrigation fraction *αirr* = *Aai1*/*Aai2* differs if the recent total irrigated area (*Aai2*) is estimated from the GMIAv5.0 data [32] (11,046 km<sup>2</sup> for the regional catchment; base case scenario), or from the national reported data [30] (13,963 km<sup>2</sup> for the regional catchment; alternative scenario).
