4.1. Neretva Valley Aquifer System
A comprehensive overview of observations taken from Neretva coastal aquifer system is shown in
Figure 5a–f, representing both rain and dry periods as well as three monitoring locations covered by the study. Following observed times series in
Figure 5, a periodic nature of confined aquifer h time series, similar to those of sea level ones, can be observed. This feature has been previously identified and analyzed to obtain hydrogeological parameters [
50,
58,
59]. During the July 2019 (
Figure 5b), D1 piezometric head is constantly kept below sea level, with an average tidal efficiency [
58] (TE) equals to 0.63. During the first half of February 2019, the occurrence of a four-day period (February 1st–4th) of precipitation with maximum amount of 39.7 mm day
−1 increases the D1 h average value with D1 h values being even greater than sea level ones (
Figure 5a). This is explained as a consequence of increased groundwater recharge from inland caused by precipitation, which diminishes with time since precipitation occurrence. Similar findings were present in work by Kim et al. [
60].
Compared to D1, piezometer D2 located approximately 4.50 km inland from the Diga coastline is characterized by on average higher h values compared to D1 but with increased attenuation of signal amplitude corresponding to TE value of 0.45 (
Figure 5d). The increase in D4 h due to the groundwater recharge component caused by precipitation and related subsurface inflow from catchment area during the first half of February 2019 equals to 48 cm, which makes this location most sensitive to groundwater inflow component (
Figure 5e,f). By coupling observed piezometric features in D1, D2, and D4, it is obvious that the groundwater flow within the confined aquifer is directed from east and southeast towards the shoreline located to the west. This is in agreement with previously performed studies [
32]. The latter holds for both dry and rain period.
The amplitude spectral density (ASD) of D1, D2, and D4 h signals in
Figure 6a,c implies the sea level as a main driving force resulting in groundwater h oscillations as found within the confined aquifer. As evidenced in
Figure 6, the sea level trend is characterized by four dominant tidal constituents, M2 and S2 as semidiurnal, and O1 and K1 as diurnal ones. Constituent characteristic values of sea level and h time series are shown in
Table 2. Amplitudes and periods are derived from the ASD in
Figure 6a,c, while phase values are calculated for each tidal constituent from FFT results as explained in Methodology section. Both rain and dry periods show dominance of tidal constituents in groundwater piezometric head, thus emphasizing the effect of direct connection of this geological unit with the sea. There is an evident dominance of low-frequency components present during rain period (
Figure 6a), which is an effect caused by both trends and meteorological changes, especially precipitation followed by the decrease of barometric pressure. The latter is of minor impact in groundwater piezometric head time series recorded during dry period (July 2019).
While the decrease of the TE is visible through ASD components as the distance onshore increases, time lag values are obtained based on cross-correlation analysis. Prior to analysis, signals are detrended by application of Fourier magnitude filter, which enables the removal of components characterized by periods higher than 44 h. The latter ensures the removal of non-tidal cause periodicity like barometric pressure changes, air temperature change caused features, etc. Cross-correlation functions are shown in
Figure 6b,d for both analyzed periods. Results demonstrate a significant correlation between sea level and h time series in confined aquifer with values decreasing with distance increase from shoreline. Maximum obtained cross-correlation value equals to 0.93 and corresponds to sea level and D1 h observed during July for time lag of 1 h. During the rain period, the maximum value of cross-correlation function is reduced to 0.90 with the same time lag value. Compared to D1, piezometers D2 and D4 are shown to be more influenced by components different than diurnal and semidiurnal tidal ones during dry and rain period. During July 2019, maximum D2 h cross-correlation function value is found to be 0.88 for 2-h time lag relative to the observed sea level time series. During rain period, cross-correlation function value is reduced to 0.81, still demonstrating significant influence of the sea level to groundwater h. The highest difference in cross-correlation values between dry and rain periods are found for D4 piezometer. While cross-correlation function corresponds to 0.802 in July, it is decreased to 0.70 in February as well as the time lag change, which has been decreased from 5 to 4 h.
Unconfined aquifer observations offer an insight to groundwater level h, EC, and T as shown in
Figure 5. Independently of the piezometer location and the period when observation has been done, groundwater level h is lower compared to one characterizing confined aquifer. This is caused by the artificial water drainage/management operated by the pumping stations whose intention is to keep groundwater low ensuring better conditions for agricultural production all over the study area. Apart from confined aquifer, h value in unconfined aquifer is significantly influenced by external loadings, such as precipitation, and shows dominant effect of PS regimes at the same time.
Although P1 piezometer is located only 40 m from the melioration channel, the dominance of the sea water and its influence on groundwater parameters is evidenced in
Figure 5a,b. Independently of the season, no significant changes in groundwater h even during intensive precipitation have been detected, keeping the presence of neap and spring tides within the P1 h time series. EC value changes slightly over the periods of interest.
During February 2019, EC took almost constant value of 50.89 mS cm
−1 while a slight increase is observed in June (51.88 mS cm
−1). Since the groundwater h value is lower than the sea level, there is an active sea water intrusion process [Bakker 2015.], resulting in P1 EC values close to EC values of the seawater. Following
Figure 7a,h shows significant cross-correlation with sea level: 0.90 in February and 0.84 in June, with time lags of 2 h. Apart from h, EC shows dependency on sea level as well, with maximum cross-correlation function value of 0.25 and 3-h time lag (February) and 0.30 and 4-h lag (July). Differences in cross-correlation values and appropriate time lags between the sea level and P1 EC values imply the balance effect between dominance of drainage channel level and of sea level. During the dry season, the pumping system works with minimum capacity due to the scarce precipitation, therefore the sea level assumes a dominant role in determining P1 groundwater characteristics. During February 2019, the system operated on a higher capacity with a total of 784 operating hours combined on all pumps, thus causing a deflection from the state found during dry season where less than 300 operating hours were recorded. Since the distance from the channel to P1 piezometer (40 m) is smaller compared to P1 and coastline distance (80 m), reduced cross-correlation value of sea level and P1 EC (0.25) as well as reduced time lag (3 h) implies the dominance of the drainage channel influence to parameters observed within P1 piezometer. T time series from P1 additionally emphasize the interconnection with sea water. In February, mean T was 12.98 °C while in July it reached 21.82 °C with a tendency of a constant increase due to the increased insolation time.
Compared to P1, time series in P2 show sensitivity to precipitation occurrence (
Figure 5c,d) through the h, EC, and T values. After 39.7 mm day
−1 precipitation rate, h was increased for 1.31 m and was followed up by a decrease in both EC and T. EC value decreased for 11.4 mS cm
−1 while groundwater temperature gauge detected a decrease of 1.8 °C. The dry period offers more stability in observed time series without significant changes. This is explained by the inspection of precipitation occurred during dry period. Precipitation absence leads to the convergence of GW parameters to stable behavior without trends. It is important to notice the EC value in July (50.9 mS cm
−1 on average) is higher than that obtained for February (24.55 mS cm
−1 on average), same year, showing the importance of precipitation occurrence. Precipitation enhances the dilution of the saline and brackish water by ensuring the volume of fresh water at the surface water bodies found within channels. This leads to long-term reduced GW salinity observed during rain period. Cross-correlation function defined for P2 piezometric head and EC in February results in maximum value of −0.88 fitting to time lag of 18 h (
Figure 7b). On the contrary, the dry period results in −0.68 cross-correlation value and only 2-h time lag. The difference in time lags implies the effects of spatially global changes during rain period and their influence on P2 EC state, while during dry period, groundwater features within P2 piezometer are dominantly defined by local conditions in the nearest area.
Piezometer P4 is located 20 m from PS Prag intake and 40 m from the Mala Neretva riverbed. As observed at P4, h shows a significant influence of PS Prag, which is mostly operated in night regime (
Figure 5e,f) which explains the daily corresponding frequency in P4 h observations. Due to the small distance from PS Prag intake, even the higher amount of precipitation does not lead to a significant h increase (
Figure 5e). Strictly, the increase in P4 h during the period February 3rd–February 4th is lower compared to P2 increase caused by same precipitation (h change equals to 0.75 m while P2 h change is 1.31 m). The same figure demonstrates that the maneuver of opening the UUU gates does not influence P4 h, although the regime of Mala Neretva river has been changed drastically as recognized via presence of tidal constituents within OUN and UUU time series. This implies that the Mala Neretva riverbed has low permeability which is ensured by the construction features of its embankment. Groundwater temperature emphasizes constant value over analyzed periods except for the local decrease of 1.9 °C after precipitation observed on 3 February. EC value shows an initial increase but after the precipitation reaches a value of 39.7 mm day
−1 it underwent a decrease of 2.74 mS cm
−1. On February 4th, after the precipitation of 39.7 mm day
−1 on the previous day, EC value shows an almost linear increase with a tendency to converge to some value of 6 mS cm
−1. The remaining period shows that the increase in h leads to EC increase and vice versa. Cross-correlation functions calculated for P4 h and EC values show maximum values of 0.57 and 35-h delay in February and 0.69 and 37-h delay in June (
Figure 7c).