*2.1. Study Area*

Spercheios river basin is located in the prefecture of Pthiotida in Central Greece, covers an area of 1,661 km2, and has a mean altitude of 641 m and a dense hydrographic network (Figure 1; [32]). The main human activities of the wider area since it was first inhabited in the Early Neolithic period [33] include arable agriculture and grazing, while industrial activities are limited mainly to small manufacturing units of agricultural products and olive oil refineries [34]. The main hydromorphological modifications of the area include water abstractions for irrigation, water flow regulations (small weirs, water distributor), canalization, and the partial diversion of the original route of the river close to its estuary. Spercheios river wider area has been included in many environmental protection networks (for example NATURA 2000, CORINE biotopes, and Wildlife Refuges; [32,35]).

**Figure 1.** Study area.

## *2.2. Hydrological Simulation*

## 2.2.1. Methodological Approach

The modelling tool used in the present study was the MIKE SHE, developed by the Danish Hydraulic Institute Water and Environment. MIKE SHE is a physically-based distributed model that is able to simulate all hydrological domains within the land phase of the hydrological cycle in a river basin. MIKE SHE is fully integrated with the channel flow code MIKE 11, which is a one-dimensional model that can simulate water flow and level, water quality and sediment transport in rivers, flood plains, irrigation canals, reservoirs, and other inland water bodies [36]. The hydrological model has already been successfully set up, calibrated and validated during a previous study for Spercheios river basin [35].

More specifically, during a previous research study, the hydrological model of Spercheios river basin was set up and calibrated for the hydrological years 2008/2009–2010/2011 and validated for the hydrological years 2013/2014–2014/2015 [35]. These periods were chosen based on the data availability (actual in situ observations of water level and discharge and high-quality climatological data) and on the fact that in 2008 the construction of the last engineering flood control structures in the hydrological network and the river banks were completed. The calibration and validation periods' length were considered to be adequate since most studies addressing the question of the utility of additional data in terms of the length of available discharge time-series in hydrological model calibration concluded that several years of data ranging between 2 and 8 years are su fficient for reliable parameter identification [37]. Moreover, when 2–3 years of continuous daily discharge data are available, then the model activates the complete set of its procedures, and the use of longer data sets would not o ffer a significant benefit in the definition of the model's uncertainty [38]. The results of the Spercheios river basin hydrological model calibration showed a satisfactory agreemen<sup>t</sup> between observed and simulated water levels and discharge measurements. Their correlation coe fficient *R* can be characterized as moderate (0.55) to high (0.77) based on the criteria for correlation interpretation proposed by Hinkle et al. [39], and in all cases the data were statistically significant at the 0.05 level, indicating the su fficient performance of the model. During validation, the resulted correlation coe fficient *R* was also moderate (0.68) to high (0.84) [39], and the data were also statistically significant at the 0.05 level. The model performance can be considered satisfactory since the results meet the criteria proposed by Moriasi [40] (*R2* > 0.50, *RSR* < 0.70, and *PBIAS* ± 25% for streamflow) in the cases where river discharge data were available for validation [35] (Table 1).


**Table 1.** Statistical characteristics and e fficiency criteria for the calibration and validation of the hydrological model at the Spercheios river basin [35].

\* result significant at *p* < 0.05; L: Level (m); Q: Discharge (m<sup>3</sup>/s); *N*: number of sample pairs; *ME*: mean error; *MAE*: mean absolute error; *RMSE*: root mean squared error; *R*: correlation coe fficient; *R2*: Nash-Sutcliffe coe fficient of efficiency; *PBIAS*: percent bias; *RSR*: RMSE-observations standard deviation ratio.

In order to investigate the impact of land cover change on the hydrological cycle components, the calibrated hydrological model of Spercheios river basin was integrated for 45 hydrological years (1960/61–2004/05) for three di fferent land cover case studies. The specific period was characterized by a

stable hydrographic network with minimum engineering interventions. Any hydraulic construction built after 2006 was omitted from simulation procedure, while all engineering interventions which took place before 1960 were included into the simulation, for the best representation of the actual state of the river network during the period 1960/61–2004/05. For the specific simulation period, the gridded time-series of the meteorological dataset used for the specific study area was complete and without gaps. Finally, the land cover case studies were selected according to the data availability and taking into consideration the overall anthropogenic interventions in the area, aiming at the better representation of each distinguished period. More specifically, until 1960, the major hydraulic interventions and the major agricultural reform of Greece had been completed, and ever since the area used for agricultural activities has been practically stable in the study area [41]. The first available documentation concerning the land cover distribution in Greece was from the year 1960 [42]. In 1990, the first pan-European land cover data collection was utilized based on satellite image processing (Coordination of Information on the Environment- CORINE Land Cover Programme [43]), and the most recent version is from the year 2018 [44]. Therefore, the following three land cover case studies in Spercheios river basin were implemented: (1) LC1960 based on the land cover of Spercheios river basin in 1960 (baseline), (2) LC1990 based on the land cover in 1990 (mid-period) and (3) LC2018 based on the land cover in 2018 (current state) (Figure 2).

**Figure 2.** Flowchart of the current methodological approach.
