1. Introduction
The development of modeling in surface and ground water hydrology is based on the causative relationships among the hydrological cycle, climate, geomorphology, hydrogeology, and human activities interaction on the above components [
1]. According to Biswas [
2], the complexity and particularity of integrated water resources management is attributed to 24 different issues which are included in integrated water resources management. One of the 24 issues is the coupling of surface water and groundwater modeling. The difficulty of an integrated approach of these two water resources components modeling for different temporal and spatial scales is mentioned by many scientists in literature. The dynamic processes of surface and groundwater systems are mainly influenced by the hydrological and hydrogeological conditions for integrated water resources management at watershed level [
3]. Therefore, understanding surface and groundwater processes requires the understanding of the impact of geomorphology, hydrogeology, and climate on subsurface flow systems [
4]. An integrated approach leads to conclusions related to how recharge and discharge are influenced under climate scenarios and management strategies [
5]. Moreover, surface water–groundwater interactions could provide important information for contaminated areas [
6].
High concentration values of nitrates in the saturated zone is recognized as a serious threat to groundwater quality as indicated by several studies [
7,
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18]. Groundwater contamination is classified in two main categories—non-point and point sources. Nitrate in groundwater may be derived from industrial, municipal, residential, and agricultural sources. Several studies show a significant correlation between agriculture and increased nitrate concentration in groundwater [
7,
15,
16]. Non-point agriculture sources include nitrogen fertilizers, manure application, leguminous crops, irrigation return-flows, dissolved nitrogen in precipitation, and dry deposition [
19,
20]. Groundwater nitrate contamination is considered as a major concern in rural areas, where agriculture is the main human occupation and economic activity and arises as a result of increased water demands as well as increased application of nitrogen fertilizers [
21,
22]. Nitrogen is led to the saturated zone mainly in the form of nitrate due to the denitrification process which occurs in the vadose zone [
15]. Almasri and Kaluarachchi [
23] characterize nitrate anions (
) as the most commonly encountered inorganic pollutant in groundwater resources of rural watersheds, owing to their negative electrical load and the low sorption ability of soils, and most importantly to their high mobility and water solubility. Nitrate anions are dissolved in soil water and lost through the water movement in two ways: The direct way, where leaching is performed at the vertical movement of dissolved molecules of nitrogen through the soil profile to the saturated zone; and the indirect way, where the surface runoff solutes are moving over the ground surface until they reach open channels such as field drains, rivers, or lakes [
12,
24]. Extensive agricultural land use is related to subsurface nitrate contamination that has been linked to adverse health concerns [
21,
22,
25]. There are two types of farming systems—the intensive cropping system and the extensive one. The intensive consists of the arable crops which predominate in Mediterranean areas [
26]. The leading forces of utilizing nitrogen-based fertilizers in agriculture is the globally increasing demand for food, the improvement of the crop yield as well as the quality of products. This is primarily caused by a growing world population with a high demand for food production and food quality [
27]. Groundwater contamination possibly leads to social and economic losses which may be also influenced by climate change, crop pattern, and crop management [
28].
Agriculture activity is closely linked to both ecosystems and society under a two-way relationship: it is obviously affected by environmental changes and, at the same time, its practices directly affect the environment. Climate change is a continuously variable global phenomenon effecting various factors including water quantity and quality assessment as well as surface water and groundwater systems management [
29,
30]. Human activities in ground and water resources management are, in turn, in close relationship with climate scenarios, especially within the Mediterranean region, an area recognized as one of the world’s most affected by climate change. Local climatic simulation models estimate a significant decrease in precipitation and higher temperatures in the region for the forthcoming period [
31]. Furthermore, climate variability and climate change reveal negative impacts in the hydrogeological environment with the decreased recharge rates and the increased water withdrawals affecting the renewable groundwater resources. Impacts on the saturated zone could also be attributed to human activities due to different land use and crop management practices [
32,
33]. Therefore, the determination of climate change consequences on groundwater systems requires future assessment of the hydrological variables that interact directly with the hydrogeological environment (e.g., groundwater recharge, pumping, pollutant leaching, etc.). The assessment of future conditions impacts (climate, land use, water demands, etc.) on groundwater systems, is achieved with the coupling of mathematical models which represents hydrological and hydrogeological processes [
30]. In this study, surface-groundwater integrated modeling is applied in order to investigate the impact of climate change scenarios and the effects of different operational water resources management strategies on groundwater quantity and quality at Lake Karla watershed, Thessaly, Greece. The paper emphasizes groundwater quality investigation and particularly emphasizes the simulation of advection and dispersion of nitrate mass of the Lake Karla aquifer.
3. Results
The calibration results for surface hydrology and groundwater flow models on watershed level have been presented in the previous studies [
36,
40]. Well-known and widely used statistical parameters have been used to evaluate the goodness-of-fit of the applied hydrologic models for calibration and validation periods [
15,
60]. Goodness-of-fit parameters are the Nash–Sutcliffe Model Efficiency (Ef) (Equation (2)), the Root Mean Squared Error (RMSE) (Equation (3)), and the Coefficient of Determination (R
2).
where,
is the mean observed value,
Ym is the simulated value, and
Yo is the observed one at time
t.
where,
Ym is the simulated value,
Yo is the observed one, i is the code number of observation points, and n is the number of observation points.
The UTHBAL model was calibrated with the observed sporadic monthly values of Lake Karla’s basin runoff to the Pagasitikos Gulf for the historical period of October 1960 to September 2002, using the multistart Generalized Reduced Gradient algorithm and the split sample test. The Nash–Sutcliffe Model Efficiency was equal to 0.66 for development (using 2/3 of the data selected with random sampling) and validation (using the rest 1/3 of the available streamflow data) periods. The LAK3 model was calibrated against the reservoir’s water level results of the UTHRL model, under the full operation hypothesis. This is due to the absence of historical observed data of the reservoir’s water stages, since the reservoir is still not fully operated in the benefit of the agricultural demands. The Coefficient of Determination (R2) of LAK3 calibration was equal to 0.7805. For the used models UTHRL and LAK3, no validation process was conducted since the reservoir is not in operation yet and no observations are available for validation purposes. The MODFLOW model was calibrated for spatially distributed hydraulic conductivity against 27 observed hydraulic heads for the period of 1997–1987. RMSE was 1.252 and R2 value was 0.989 for the calibration period. The validation period of MODFLOW occurred on May 2002 using 10 wells for the assessment and four statistical criteria (Mean Absolute Error (MAE) = 1.003, RMSE = 1.65, γ = 0.985, and R2 = 0.993). Hence, only the simulation procedure of advection and dispersion of nitrates is presented using the integrated framework to assess natural and man-made effects on groundwater quality. Finally, it is worth mentioning that the dry and wet periods that were used had been characterized on the available historical meteorological data.
3.1. Nitrate Transport and Dispersion Model Calibration
The model calibration procedure is followed in order to assert that the simulated nitrate concentration values are close to the observed ones. The model was calibrated considering the nitrate leaching parameter as the guidance criterion, using the trial and error approach for the observed nitrate concentration data values on 1 September 2007. The visual inspection of
Figure 5 (R
2 = 0.96) and the performance measure explained by the Nash–Sutcliffe model efficiency coefficient (Ef = 0.95) demonstrated the successful calibration process. The validation process of MT3DMS took place on September 2008 and 22 wells were used. Moreover, the statistical parameter R
2 was used for the validation period and it was equal to 0.83.
Table 2 shows the starting and final parameter values before-and-after the calibration process of the most significant crops.
3.2. Operation Strategies Results
The operational strategies were performed for the historical period 1995–2007. In these strategies no water demand sub-scenarios were examined. The results regarding to the hydrological modeling on watershed level, were performed by the UTHBAL, UTHRL and LAK3 models and described in the paper of Tzabiras et al. [
40]. As expected, between the two scenarios, there was no difference in UTHBAL results as the climatic conditions were the same. Furthermore, there were no UTHRL and LAK3 results for the non-reservoir scenario. However, for the reservoir scenario, UTHRL resulted in a positive volumetric water budget of the aquifer and LAK3 calculated the annual seepage volume at 18 hm
3 into the aquifer. These results are fundamental and used as input data of the groundwater nitrate contamination modeling system.
3.2.1. Groundwater Hydrological Modeling
Annually observed water withdrawals of 80 hm
3 from the non-renewable resources of the subsurface water system for the no-reservoir scenario described the aquifer’s over-exploitation. As a result, the hydraulic head reached the value of −80 m at the south part of the aquifer area, which is depicted in
Figure 6a. On the reservoir scenario implementation, although 62 hm
3 per year were exploited from the non-renewable resources, there was a significant rehabilitation of aquifer’s water table at nearby reservoir areas including the southern area of the aquifer (
Figure 6b).
3.2.2. Nitrate Transport and Dispersion Modelling
The spatial distribution of nitrate concentration values regarding the no-reservoir scenario, was modelled and presented in
Figure 7a. The nitrate values were significantly increased, reaching 33 mg/L in the south-western and south-eastern part of the study area. The large nitrate concentration values were attributed to the excessive application of nitrogenous fertilizers and to the prevailing crop pattern on the study area. According to the reservoir scenario, the results of nitrate concentration values were considerably lower than the no-reservoir scenario. The reservoir operation scenario signaled the increased recharge which is also explained by the direct hydraulic connection between the reservoir and the aquifer. The spatial distribution of its nitrate concentration values is presented in
Figure 7b. The maximum nitrate concentration value was equal to 24 mg/L and the lower ones were mostly observed in the south-eastern part of the aquifer, where the reservoir is located. Hence, as nitrates are water-soluble contaminants and the recharge rate is considerably increased, the combination of these two factors resulted in the decrease and dissolution of nitrates in the aquifer. Comparatively evaluating the two management scenarios, the percentages of differences of nitrate concentrations in the surrounding area of the reservoir, ranged between 10% and 15%. Thus, it can be concluded that the contribution of the reservoir operation is of critical importance towards the remediation of the groundwater quality.
The maximum allowable nitrate concentration limit according to Directive 98/83/EC is the value of 50 mg/L, while the upper limit of 25 mg/L is also defined as the ‘‘guidance value’’, above which the water is characterized as not appropriate for domestic use, since above the defined nitrate upper threshold, severe health issues are caused. The nitrate transport modeling of the no-reservoir scenario showed gradually increasing concentrations of the nitrate compounds in the aquifer, in analogy to the fertilizer application rates on the studied area. The presence of the reservoir, conversely, leads to the increased reservoir’s recharge and the decrease of nitrate concentrations in the groundwater system. Furthermore, the reservoir operation leads to local decrease of nitrate concentrations in the surrounding areas, even below the upper indicative threshold of 25 mg/L, as shown in
Figure 8a. Regions with nitrate concentrations above the indicative threshold of 25 mg/L for both the no-reservoir and the reservoir scenario, respectively, are depicted in
Figure 8b. As a result, for the inversion of this environmental degradation and the prevention of water pollution, radical measures should be taken for the study area, following the application of good agricultural practices, and the adoption of efficient irrigation methods for the no-reservoir scenario.
3.3. Climate Change Results
3.3.1. Surface Hydrological Modeling
Groundwater recharge is the most essential hydrological component factor that affects nitrate leaching in the aquifer. The annual rainfall decrease observed in all three socioeconomic climatic scenarios has direct impact on the groundwater recharge rates. Annual average recharge rate was estimated to be 41.7 hm
3 for the historical period. For the mid-term period of 2030–2050, annual recharge rate is estimated to be 40.7 hm
3 for the SRESB1 scenario, 40.5 hm
3 for the SRESA1B scenario, and 39.1 hm
3 for the most adverse scenario SRESA2. For the long-term period of 2080–2100, it is estimated to be 40.2 hm
3 for the SRESB1 scenario, 37.9 hm
3 for the SRESA1B scenario, and 37.3 hm
3 for the SRESA2 scenario [
40].
3.3.2. Groundwater Hydrological Modeling
The overall simulation results highlight the impact of climate change and the effectiveness of the three sub-scenarios of two water resources strategies on the groundwater quantity regime, as shown in
Table 3 and
Table 4. More specifically, applying of the two water resources management scenarios, the sub-scenarios, and the socioeconomic storylines for the mid and long-term period, the average hydraulic head drawdown was estimated at about 200 m. The largest drawdown (210.73 m) was observed in the mid-term period of the SRESA1 socioeconomic scenario. For the SRESB1 scenario, the drawdown reached the value of 231.40 m for the mid and long-term periods.
3.3.3. Nitrate Solute and Transport Model
The nitrate concentration results which were estimated for the two operational strategies indicated significant differences. Conversely, the resulting differences were minimized during the application of the three socio-economic storylines. The groundwater nitrate contamination results of the two operational scenarios for the historical period and for the three socioeconomic storylines are presented in
Table 5 and
Table 6.
The maximum nitrate concentration value range increased from 31.87 mg/L in the mid-term period, to about 50 mg/L in the long-term period due to climate change without antropogenic activities (no reservoir operation). The highest values of nitrate concentrations was recorded to be 50.48 mg/L in the long-term period of the SRESA2 scenario. Furthermore, the highest values of nitrate concentration were observed at the south-eastern region and portrayed in
Figure 9. Additionally, the lowest values of nitrate concentrations were recorded in the mid-term period of the SRESA1B scenario in the north-eastern and south-western part of the study area and are also presented in
Figure 9. On the contrary, the nitrate concentration values ranged from 0.20 to 32.64 mg/L, in the historical period. The maximum nitrate concentration value of 32.64 mg/L was recorded in the southeastern area while the minimum value was encountered in the northern part of the study area.
Spatial nitrate concentrations that exceed the indicative threshold value of 25 mg/L regarding the mid-term and long-term period, are depicted in
Figure 10a,b, respectively. The comparison of
Figure 8 and
Figure 10 reveal the impact of climate change on nitrate concentrations, since the areas where the value of 25 mg/L is exceeded for the two future periods are significantly larger than the one for the historical period. The percent differences between the three socio-economic plots and the historical period range between 0% and 15% for the mid-term period, and between 0% and 30% for the long-term. Their spatial distribution is illustrated in
Figure 11. The differences for the historical period of the reservoir scenario are not depicted, because in neither of the operational nor the climate scenarios was the indicative value of 25 mg/L exceeded. The groundwater quality spatial status for the two water resources operational scenarios in conjunction with the climate change storylines is shown in
Figure 9. The above results highlight the necessity of continuous monitoring of groundwater quality in the study area, since many wells are used to supply water for domestic and irrigation uses in the surrounding settlements and in the city of Volos [
34].
4. Discussion
One of the most important impacts of climate change and variability on water resources is the reduction of groundwater recharge rate and aquifers’ water table depletion [
17,
32,
33]. This fact affects groundwater quality, leading to the increase of nitrates concentration [
17,
31,
61]. In our research, for the long-term period of SREA2 scenario without reservoir operation, groundwater recharge was reduced to 4.4 hm
3 per year and hydraulic head drawdown reached 45 m resulting to about 18 mg/L increase of nitrates concentration. Furthermore, the area of 25 mg/L nitrate concentrations exceedance of aquifer was expanded due to the climate change impact.
On the contrary, the water storage infrastructures operation contributes to the groundwater recharge raise due to their hydraulic connection with the underneath aquifer [
62,
63]. In the cases where these infrastructures cover the irrigation needs of the nearby cultivations, a significant rehabilitation of aquifer water table was observed [
64,
65]. The results of this study highlight this benefit. Reservoir operation, for the historical period, increased the groundwater recharge to 18 hm
3 per year. Therefore, an elevation of the hydraulic heads at 40 m leading to a reduction of 20 mg/L nitrates concentration was observed in the reservoir vicinity. This concluded to a 41% area reduction of 25 mg/L exceedance.
The land use and the hydrological and hydrogeological data used in this study, refer to the historical period 1995–2007. Newer land use data have been compared with the relevant data of 2001 and the changes were found to be marginal. On the other hand, newer datasets of hydrological and groundwater data are not available. All the component models of the modelling system were calibrated successfully, indicating that the models were able to simulate accurately the hydrological and hydrogeological processes of the watershed and the aquifer of the study area. The calibrated models were also able to capture the future hydrological response because there were no significant changes in the state parameters of the surface and groundwater hydrological processes. Therefore, the use of older datasets was acceptable. They did not introduce additional uncertainties in the results of the paper, and they did not affect the conclusions of the paper.
5. Conclusions
The present study described and evaluated the applicability of an integrated modeling system approach on water management strategies and climate change impacts assessment on the water resources quantity and quality of the Lake Karla aquifer. Nitrates are the most important pollutant factor and nitrate pollution is a known global issue according to worldwide scientific literature, which is caused by the need for increased food-supply production to address the needs of the growing world population. The modeling system included coupled surface, groundwater, and reservoir’s operation hydrological simulation tools that clearly focused on the advection and dispersion of nitrate mass in the groundwater regime.
The groundwater quantity and quality regimes were examined on Municipal District level. Overall, six original water management strategies were compared and evaluated. Two management strategies described the aquifer’s status, one without taking under consideration the operation of the reservoir (no-reservoir scenario) and one taking it under consideration (reservoir scenario). Both cases were sub-linked to three water demand strategies. The climate change impacts were also evaluated for three socio-economic storylines. The applied water resources management strategies’ simulation results under climate change scenarios were milder for the mid-term period, as compared to ones for the long-term period that were more moderate as impact results. Regarding the nitrate contamination of groundwater resources, the management strategies and the current crop pattern applied continued to have significant negative impacts on the groundwater quality regime.
Based on the simulation results, it is concluded that the current agricultural activities were the main reason for the groundwater quality degradation. This phenomenon becomes more intense with climate change impacts. Operational management practices in conjunction with water demands methods compensated the groundwater resources nitrate contamination of climate change impacts. Although the reservoir’s operation in the study area is evidently an efficient groundwater resources management tool, promoting not only the sustainability of the general water regime, but also the regional socio-economic development, this approach of groundwater resources recovery alone, is not a panacea. The adoption and implementation of good agricultural practices in conjunction with the application of appropriate crop pattern, that comply with the objectives of water resources quality remediation constitutes a great necessity.