3.1. Extension of Saline Water Intrusions along the Central Axis of the VRE
Sections of salinity along the VRE during spring tide sampled in austral spring 2017, austral summer 2018, austral fall 2018, and austral winter 2018 are displayed in
Figure 2 (to identify the locations of the individual stations please refer to
Figure 1). Each section was sampled during ebb and flood tides from the innermost to the outermost station; both sections were averaged. The combined river discharges (7 days prior to sampling) were different in all seasons sampled: from austral spring to winter 464, 252, 332, and 785 m
3 s
−1. The intrusion of saltwater (a threshold of 1 PSU was chosen) was indirectly proportional to the strength of river flow, i.e., the lesser the flow, the farther the intrusion extended. In austral summer, on 1 February 2018, when the river discharges were at their minimum, saltwater was detected 22 km upstream from the estuary entrance. By contrast, in austral winter (12 August 2018), when the river flow was high, saltwater reached only 15 km into the Valdivia River. During austral spring and fall, both flow rate and intrusion were at intermediate levels. Saltwater intruded somewhat farther into the river along the bottom in all sections, and in that position there was little stratification of the water column.
In order to statistically relate the extension of the saltwater intrusions to the river discharges of the Calle-Calle and Cruces rivers, linear, power, and exponential regression analyses were carried out considering five discharge measures (see Data and Methods). Saline intrusion estimates were obtained from the seasonal hydrographic sections during spring and neap tides (
Table 1); ebb and flood tide hydrographic transects were averaged. The regression results are presented in
Table 2 for all combinations of regression models and discharge measures. The
R2 values ranged from 0.53 to 0.88. The power regression models, in general, yielded the best results, followed by the exponential and linear models. The discharge measures, complete hydrologic flood, complete value, and complete averaging methods, which included the discharge of both rivers, generated higher correlations in all regression models. In all regression models, the average flow of the sum of both rivers on the seventh day before sampling (complete value method) yielded the highest
R2 values: 0.88, 0.77, and 0.72 for the power, exponential, and linear regression models, respectively.
The curves of the power regression model used to fit saltwater intrusion to flow rates by applying different methods to estimate the flow are shown in
Figure 3a. When discharges were low (<600 m
3 s
−1), the flow estimates for the different methods used were not widely dispersed and in general the model fit to determine saltwater intrusion was good. However, when discharges increased, the dispersion of flows according to different measures was wide. This was especially noticeable for the austral winter–spring tide sample (second to last seawater intrusion), where for example for the complete value method (7 days prior to sample) and complete hydrologic flood method (maximum flow rate within a 15-day window prior to sampling) the flow values were 786 and 2502 m
3 s
−1, respectively. Persistent high flow rates would, however, not allow a saltwater intrusion of 13 km (see
Section 3.3), and in this case R
2 is only 0.7; thus the complete hydrologic flood method was rejected. The power regression model in which the complete value method discharge measure was applied yielded the best fit and explained 88% of the observed variance: intrusion (km) = 163.6; discharges (m
3 s
−1)
−1/2.64; it was thus chosen based on the highest squared correlation.
During the particular times in which the hydrographic transects were performed, the combined river discharges 7 days prior to sampling happened to be in the range of 252–786 m
3 s
−1. Long-term (2008–2019) seasonal averages of river discharges were for January–March, austral summer (198 m
3 s
−1), April–June, austral autumn (470 m
3 s
−1), July–September, austral winter (957 m
3 s
−1), and October–December, austral spring (465 m
3 s
−1), which puts our seasonal snapshots (
Figure 2) in the range of average or less than average conditions. When applying the best regression fit to these flow rates, saltwater intrusion resulted to be between 12 (austral winter) and 22 (austral summer) km (
Figure 3b), which was in good agreement with the in situ observations, although the austral winter cruise registered an intrusion of 15 km due to the lesser than average long-term flow rate. The austral summer cruise did not coincide with annual minimum river discharges, which usually occurred by the end of March (see
Section 3.3); therefore, intrusions are expected to reach beyond those 22 km as estimated for 1 February 2018. Between 2008 and 2019, minimal (maximal) discharges were lesser (higher) than 60 (3800) m
3 s
−1 on 28 March 2015 (1 September 2008) with an estimated saltwater intrusion of 35 (7) km (
Figure 3b).
3.2. Harmonic Analysis of Tides
To bridge the gap between the seasonal hydrographic samplings and to enhance the time of observations, a mooring was installed just south of Valdivia City, labeled here Lido station; surface and bottom salinity and river level were registered from November 2017 to May 2019, and in Valdivia station somewhat farther downstream on the Valdivia river from November 2017 to July 2018 (see Data and Methods). The main harmonic constituents that made up the astronomical tidal wave at the Lido and Valdivia stations were the M2, K1, S2, O1, and N2 constituents from highest to lesser significance, and they explained 85.75% and 96.81% of the observed water level time series, respectively (
Table 3). The amplitudes of the main components of the tides were similar in both stations, adding up to 0.74 and 0.77 m for the Lido and Valdivia stations, respectively. The astronomical reconstructions of both water level records were out of phase by 20 min (Sampling interval was 10 min) and arrived first at Station Valdivia as expected. The sum of the amplitudes of the M2 and S2 constituents, which determine the bi-weekly variability in water level height (14.8 days, spring–neap tide cycle) were more important than the sum of the amplitudes of the M2 and N2 constituents, which define the 27.6-day perigee–apogee cycle in water level height.
The factor
F relates the sum of the amplitudes of the two most important principal diurnal harmonic constituents (K1 and O1) to the two main principal semi-diurnal ones (M2 and S2). For the Lido and Valdivia stations,
F resulted in 0.51 and 0.48, respectively, thus classifying both tidal records as mixed tides predominantly of the semi-diurnal type [
27].
3.3. Sub-Tidal Variability at Station Lido
The impact of tidal range variability due to spring–neap and perigee–apogee cycles, plus combined low river flow on salinity intrusion and stratification, is shown in
Figure 4. This figure illustrates the impact of tidal range variability owing to spring–neap and perigee–apogee cycles and combined low river flow on salinity intrusion and stratification. By far the largest contributor to river flow in the VRE is the Calle-Calle River, which surpasses the Cruces River’s contribution by about seven times (
Figure 4a). The river’s discharge exhibited a seasonal distinction, with the highest flows from April to November 2018, reaching maximum runoffs in August 2018 (>2200 m
3 s
−1) and the lowest between Decembers to May 2017–2018 and 2018–2019 (<500 m
3 s
−1). A year-to-year difference could be observed, with a lower average flow between November–May in 2018–2019 compared to 2017–2018. The tidal range properly preserved the fluctuations typical for bi-weekly and monthly variability, namely spring–neap tide and perigee–apogee tide (
Figure 4b,c). Saltwater intrusions of up to 5 PSU took place during both periods of low river discharges, in November–May, and only then, but were more intense and longer-lasting during 2018–2019 than during 2017–2018, consistent with the lower river flow in the former period. In general, these saline water intrusions occurred when the total river flow was less than 280–300 m
3 s
−1 and were approximately in phase with the spring and neap tides, with greater increases in salinity and stratification during spring tides. Modulations of monthly variability were best observed in the 2018–2019 period, which was longer. Therefore, it was observed that both cycles, spring–neap and perigee–apogee, modified salinity when river discharges were low, producing higher salinities during spring tides.
The cross wavelet transformation measures the similarity between two time series, here the tidal range and bottom salinity. Common power during the time of observation is shown in color in
Figure 4d. The phase relation between both time series during the time is indicated by the direction of the arrows. The only periods of high common power were from January to May, the times of the lowest river discharges in 2018 and 2019 as well. High similarity in power during these periods was centered on the biweekly and monthly periods, mostly in phase (arrows pointing to the right). The signal was stronger in the bi-weekly period in both occasions. Common high power was observed for a longer time in 2019, in agreement with a longer phase of low river flow.
3.4. River Discharge and Intra-Tidal Constituents Control Salinity at the Lido and Valdivia Stations
Hourly time series of the combined discharges of the Calle-Calle and Cruces rivers, the reconstructed tidal record, surface and bottom salinity at the Lido station, and bottom salinity at the Valdivia station, all for March and April 2018, are depicted in
Figure 5. The control that the height and phase of the tides have on salinity during conditions of low river discharge is effectively revealed in this particular time window. During times of low river discharges, around 150 m
3 s
−1, during the first 18 days of March (
Figure 5a), a well-established synchrony between the height of high and low tides and salinity (bottom and surface as well) was observed (
Figure 5b,c). Salinity oscillated with the tides between 1–12 PSU, where the bottom salinities exceeded those of the surface, thus giving rise to periodically occurring salinity intrusions related directly to tidal periods. At the Valdivia station, which is situated somewhat closer (11 km from hydrographic station 4,
Figure 1) to Corral Bay, salinities even reached more than 15 PSU close to high tides, although this record presented many gaps due to limitations of the conductivity sensors’ range (see Data and Methods). When the combined discharges more than doubled abruptly, on March 17, to values above 400 m
3 s
−1, the salinity fluctuations related to the height of the tides drastically reduced, to values of less than 0.5 PSU at the Lido station. The combined river flow dropped again between 21 March and 10 April; however, it did not drop to the levels of beginning of March but nevertheless to around 300 m
3 s
−1. Only from time to time a few weak bottom salinity peaks were registered, of up to 2 (10) PSU at the Lido and Valdivia stations. The magnitude of these minor peaks increased around 10 April to the neap tide levels of the beginning of March. River discharges then increased with the onset of austral fall to well above 400 m
3 s
−1, thus shutting down saltwater intrusions at its part of the VRE.
The effect of the tide on the saline intrusions at the Lido and Valdivia stations could be better understood by carrying out a more detailed analysis between 13 and 16 March 2018, during nearly spring tide with river discharges of less than 200 m3 s−1. Surface salinity fluctuations were in phase with the height of the tides, with values ranging between about 0–8 PSU at the Lido station. However, at the bottom of the river maximum salinities occurred about an hour later during flood tides with slightly higher salinity, up to more than 10 PSU at Valdivia station. During flood tides, this delay caused the stratification measure (bottom–surface salinity) to increase and reached values above 2 at the Lido station, whereas the stratification decreased during ebb tides (values close to 0).