Sea-level trends along freshwater and seawater mixing in the Uruguayan Rio de la Plata estuary and Atlantic Ocean coast

Sea level is rising worldwide with local differences due to global and regional drivers. This article analyses yearly freshwater and sea level trends and fluctuations during the mixing of freshand sea-water along the Uruguayan coast of the Rio de la Plata River estuary and the Atlantic coast from 1961 to 2014. The global and regional drivers as well as local co-variables are described, classified in nine discrete classes and inter-correlated. Despite the observed increasing trends, local sea level rises (SLR) are not well correlated with global SLR except at the estuarine-ocean boundary (Punta del Este station). Freshwater inflow, which variability often coincides with Oceanic El Niño-La Niña (ONI-ENSO) events, is the first descriptor of sea level fluctuations and outliers all along the coast, particularly at Punta del Este. Local SLR roughly follows the overall global trend with periods of acceleration and stabilization often coinciding with ENSO events.


Introduction
Sea level rise (SLR) is changing the dynamics at play along coasts. The dynamics at play includes (UCS, 2013):  Amplified wind-storm surge. With rising seas, storm surge occurs on top of an elevated water level (UCS, 2013).  More intensive shoreline erosion, degradation and coastal destabilization. SLR increases the potential for erosion by allowing waves to penetrate further inland, even during calm weather (Bruun, 1962;Shepard et al., 1962;Zhang et al., 2004;Syvitski et al., 2005).  Permanent inundation of low-lying coastal lands (Cooper et al., 2008).
Because of tidal and wind-driven changes, sea level is constantly fluctuating. Therefore, it is important to calculate the mean sea level (MSL), which is the average sea level at a given location over several years (Douglas, 2001). According to Chao et al., (2002) sea level change occurs on all time-scales, and with a continuous range of spatial scales, from local (e.g., wind storm surge), to regional (e.g., El Niño Southern Oscillation (ENSO)), and global (eustatic). On decadal to multi centennial time scales sea level fluctuations are mainly driven by climate change in response to natural forcing factors (e.g. solar radiation variations, volcanic eruptions) and to internal variability of the climate system (related for example to atmosphere-ocean perturbations such as El Nino-Southern Oscillation -ENSO, North Atlantic Oscillation -NAO, Pacific Decadal Oscillation -PDO) (Meyssignac and Cazenave, 2012).
Most estuaries have a series of landscape subcomponents: i) a fresh water source, ii) a tidal-estuarine segment, and iii) a pass to the sea. The interaction of three primary natural forces causes estuaries to be unique and different (Montagna et al., 2013):


Climate: causing variability in the freshwater runoff, which is fundamental to the functioning of estuaries.  Continental geology and geomorphology: causing variability in elevation, drainage patterns, landscapes, and seascapes (Measurements of isostatic land uplift or downlift are not available yet and is assumed not to be relevant in coastal Uruguay).  Tidal regime: causing differences in the degree of mixing and elevation of the mixing zone.
This study review the background and investigates sea level (SL) trend patterns and fluctuations at four stations along the freshwater and ocean mixing along the Uruguayan coasts of RdlP river estuary (Argentina-Uruguay) and the Atlantic Ocean over the last 50 to 60 years. Emphasis is put on the relationships of SL with global and regional drivers of trends, fluctuations and extremes, including a few local co-variables from 1961-2014.
2 Study Area and Background 2.1 Geographical setting The Uruguayan coast is 670 km in length with 450 km lying within the RdlP estuary and the remaining 220 km on the Atlantic Ocean ( Figure 1). Figure 1 Rio de la Plata basin and river estuary, Southeastern South America. The four tide gauges are shown (red circles), and weather stations Carrasco and Laguna del Sauce airports (violet circles). The turbidity satellite image shows the divide of tidal fresh turbid water and estuarine marine water for a very low La Niña-linked river inflow. Modified from Nagy et al. (2014a).
International Journal of Marine Science, 2016, Vol.6, No.7, 1-18 http://ijms.biopublisher.ca 3 Figure 2 Río de la Plata bathymetry and morphological units. The most relevant for the hydology and water/sea level are the natural channels along the northern coast (Sistema Fluvial Norte, Canal Norte and Canal Oriental) which canalize fresh-and estuarine discharge, and. the "estuarine delta" (Barra del Indio) which roughly represent the upward limit of salt intrusion. Source: López Laborde and Nagy (1999) The RdlP is a large (38 x 10 3 km 2 ; 40-30 km wide) river-influenced tidal river and primary estuary (Perillo, 1995a,b), defined by López Laborde and Nagy (1999) as "a funnel-shaped coastal plain tidal river with a semi enclosed shelf area at the mouth and a river paleovalley at the northern coast that favors river discharge and sediment transport to the adjacent continental shelf" (Figure 2).

Hydrological and climatic background
The main system's forcings are the big tributary discharges (namely Parana river and Uruguay river), the weak tidal currents (amplitude < 0.5 m) and front that propagates from the sea, and the action of winds (Balay, 1961;Guerrero et al., 1997;Simionato et al., 2007;Nagy et al., 2008a;Meccia et al., 2009).
The tide wave comes from the Atlantic Ocean, being deformed within the estuary due to the shape, banks, channels, depth, and Coriolis deflection towards Montevideo, where the cross-sectional area sharply decreases ( Figure 3). Therefore, isoamplitudes increase upstream the transverse sections 24-20 (estuarine front). This facilitates the occurrence of "storm surges", that is to say the positive anomaly between the astronomic tide and the observed water level due to residual effects of winds, waves, sea level pressure (SLP), and freshwater inflow (Balay, 1961;Nagy et al., 1997;López Laborde and Nagy, 1999;Luz Clara, 2014;Verocai et al., 2015).
The fluctuations of freshwater inflow and axial offshore and onshore winds influence those of sea-level (Balay, 1961;Verocai et al., 2015). For this reason, the observed increase of river flow (García and Vargas, 1998;Nagy et al., 2002;2014b) and Southeastern (SE) and East-Southeastern (ESE) winds over the last few decades (Escobar et al., 2004), which are the main causes of storm surges in the RdlP (Balay, 1961;D'Onofrio et al., 2008;Verocai et al., 2015), is changing the balance of the physical forcings on the system. Historically, as a consequence of the permanent South American anticyclone (High Pressure Belt) over southeastern South America (SESA), the predominant wind direction has been from Northern quadrants (N to NE) (Balay, 1961;Nagy et al., 1997). An increase of S, SE and ESE wind directions was reported over the RdlP region since the 1960s (Escobar et al., 2004;Simionato et al., 2005;Nagy et al., 2008a, b;Meccia et al., 2009;Ortega et al., 2013;Pescio, 2015) and SE wind became the predominant one in Montevideo since 1998 (Nagy et al., 2014 b;Verocai et al., 2015). A relationship between stronger E to SE wind waves at the estuarine front was reported by Panario et al., (2008), whereas an increase of ESE winds was found during moderate to strong El Niño years (Gutierrez et al., 2016). Station 3: Punta del Este, located in the estuarine-marine boundary close to the mouth of the Canal Oriental to the Atlantic ocean. There are data available from SOHMA and MTOP from 1964 to 2014. The latter, with fewer gaps (N = 42) is used.

Materials and Methods
Station 4: La Paloma, located at the Atlantic coast 80 km to the East of the mouth of the RdlP, from 1955 to 2014, with gaps (N = 51) shows. Data was available from SOHMA and the Ministry of the Environment (DINAGUA). Table 1 shows relevant water heights at station 2.

Analyzed Variables and Methods
Time-series of aggregated yearly averages of selected global drivers (ENSO-ONI Index, sea level rise), regional drivers [(Paraná river flow (Q P ), Uruguay river flow (Q UY ), freshwater inflow to the RdlP system (Q F : Q P + Q UY )], and local co-variables [(sea level pressure-SLP, air temperature-T MV at Montevideo, water level at Colonia (WL CL ), sea level (SL) at Montevideo (SL MV ), Punta del Este (SL PE ) and La Paloma (SL LP )] are presented from 1961 to 2014.
The observed continuous values were converted into discrete nine unequal classes or "class variables" (Pasta, 2009;Hauke and Kossowski, 2011) relative distribution for each one where 1 is the observed minimum minimorum, 5 the average and 9 the observed maximum maximorum (Table 2).  The discretization of data was done because some variables are not normally distributed, the relationship between them is not linear, and, particularly, some have strong outliers; the selection of a unequal 9-class sub-classification instead of a 5-class which is usually recommended for the number of data (N= 54) (Cochran, 1968;Pasta, 2009) was done both to reduce the effect of strong outliers and to represent all relevant intervals. Then a matrix of discrete Spearman's Rank-Order Correlation (r S ) of the nine classes was performed (Table 3) and the correlations with significances p < 0.05 (light gray) and p < 0.01 (dark gray) are shown. The strength of the correlation is described using the following guide for the absolute value of r S : 0.20-0.39 "weak", 0.40-0.59 "moderate", 0.60-0.79 "strong", 0.80-1.0 "very strong".
For each station yearly water/sea level (WL/SL) averages were calculated (N = 8,800 hourly data per year) referred to the former customary national zero scale (Wharton hydrometric Zero reference plane, from now on 0-Wharton). The average, linear regression and Pearson correlation (r P ) were calculated for the time-series. For ENSO years the water level anomaly (residue) was calculated for each station, that is to say the difference between the local mean level (WL/SL) and the yearly level for each year identified as moderate and strong. Then the average of these residues for the moderate, strong and all El Niño years are shown for the four studied sites.
The Oceanic Niño Index-ONI El Niño and la Niña neutral, moderate and strong events were identified (including very strong El Niño) [3 month running mean of ERSST.v4  SST anomalies in the Niño 3.4 region (5 o N-5 o S, 120 o -170 o W)], centered 30-year base periods updated every 5 years (NOAA, 2015). ONI 3-month running mean (December to February -October to December) were aggregated on yearly basis. Therefore, following NOAA-ONI ranges, eight ENSO classes are defined which adapted for the 9-classes distribution includes classes 2 to 9. ONI class 1 is unfilled because it was not recorded any La Niña event < -2ºC or equivalent to extreme El Niño (class 9). Then, only for ONI index, the minimum minimorum becomes class 2.

Oceanic Niño Index, freshwater inflow, sea level pressure, and winds
The two global drivers, SLR and ONI have shown a gradual increase. The former was likely greater than 3.2 mm per year between 1993 and 2010 (Church et al., 2013). Figure 4 shows the discretized 9-class global sea level time-series from 1961-2014.
Yearly freshwater inflow (Q F ) to the RdlP has steadily increased from early 1970s up to mid 1990s with negative and positive fluctuations during La Niña and El Niño years respectively. Over the last decade (2004-2013) six years in ten were below the central class (5), four of which classified as low discharge (3). Figure 6 shows the polynomial evolution of Q F .

Punta del Este (SL PE )
The correlation between available yearly means SL PE and time 42 years) was positive and strong (R 2 = 0,54, p< 0.0001; rate 4,7 mm/year). The SLR in Punta del Este was the highest along the Uruguayan coast, and above the global SLR rate (3.2 mm/year). During the observed period five strong and moderate El Niño events and five La Niña events coincided with recorded sea level. During El Niño events yearly SL was ≥ 92 cm (long-term SL average) and the trend-line, whereas during La Niña only two yearly SL were beneath the average, and other two were above it (2008 and 2011), but beneath the trend-line. The average anomalies (observed SL PE -average SL PE ) for strong and moderate El Niño events were -2 cm beneath 92 cm and 1.7 cm above it respectively (Table 4). The extremes occurred in 1995 (maximorum) and 1968 (minimorum).

La Paloma (SL LP )
The average SL was 89 cm and showed two relatively accelerated increases from 1955 to 1983 and from 2000 to 2014. During the observed period  15 years showed yearly SL LP higher and 9 lower than the average of 89 cm, whereas 5 El Niño events coincided with data from LP. The average anomalies for strong and moderate El Niño's were -0.60 and -0.17 cm beneath the average (  , 1955, 1971and 1989, as well as 2009when both la Niña (2008) and El Niño (2009-2010 events occurred. Table 6 shows the discrete correlation matrix (R S ) of the twelve ranked variables from 1961 to 2014 presented in Table 3. Only moderate (rs ≥ 0.4) and significant (p< 0.05 and p < 0.01) correlations were taken into account. The best correlated variable with others were Q F , and SL PE with 3 significant correlations >0.4 (N: 3), GSLR, Q P , WV MV and Q UY (N: 2), whereas ENSO-ONI, SL MV , SLP MV , WL CL , T Mv and SL LP had no significant weak to strong correlations (R S ) (N:0). Punta del Este was the only station with significant correlations (Table 6) with both global drivers (GSLR Vs SL PE , r s : 0.74 and Q F Vs SL PE , r s : 0.43). The scatter plot between the discrete series GSLR and SLR PE shows twelve years (i.e. 28%) located at more than one discrete class from the trendline (residual years) (Figure 9). In addition, the scatter plot between freshwater inflow (Q F ) and the residual GSLR-SL PE years is strongly correlated (r p : 0.76; p< 0.002). The discrete time series WV MV showed a very strong correlation with SL PE (r s : -0.88). However, the correlation between the wind discrete series with the twelve residues of the scatter plot GSLR -SL PE was weak.

Discussion
Slow onset processes like SLR will result in a range of impacts that will, to a great extent, place a disproportionately large burden on poor and vulnerable groups in developing as well as developed countries (Roberts and Andrei, 2015).
A surprising finding is that the Oceanic Niño Index (ONI) is not well and significantly correlated with any other variable, despite the observed relationships with outliers (residues) of freshwater inflow, wind regime and sea level reported in recent papers (Nagy et al., 2008 a,b;2014b;Verocai et al., 2015;Gutiérrez et al., 2016), and with values above/beneath the trendlines and outliers in the four stations.
The observed yearly sea level fluctuations along the Uruguayan coast are mostly attributable to regional and local climate patterns, e.g., freshwater inflow. Although in meso-and macro-tidal estuaries, river flow has little influence on tidal dynamics away from the upper reaches (Environment Agency, 2010), in micro-tidal environments like the RdlP, river flow has strong influence sea-ward in the lower estuary and plume beyond the geographical limits of the system, e.g. at La Paloma (Nagy et al., 2014a). There are several other periodic climate variabilities to be accounted for the regional hydro-climatologic and oceanographic fluctuations e.g., the North Atlantic Oscillation-NAO, the Atlantic Multidecadal Oscillation-AMO, and the Pacific Decadal Oscillation-PDO (Ortega et al., 2013;Nagy et al., 2014b;Gutiérrez et al., 2015) which have not been discussed here.
14 Figure 9 Scatter plots of Global Sea Level Rise (GSLR) Vs Sea Level at Punta del Este (top), and Total freshwater inflow (Q F ) Vs twelve outliers (residues) of GSLR Vs SL PE .
The most relevant finding was the strong SLR at Punta del Este, which accelerated from 1964 to 1995, matching quite well global SLR trend and regional freshwater inflow increase from about 1971 to mid 1990s. At La Paloma, SLR acceleration since 2002 follows quite well global trends. The strong correlation of global sea level rise with sea level at Punta del Este (+0.74) suggest this station is key to monitoring SLR and understanding the complex multi causal sea level fluctuations. The moderate relationship between La Niña events and negative SL residues (anomalies) over the last decade at this station is likely explained by the primary control of global sea level rise. The negative very strong correlation with wind velocity at Montevideo (-0.88) seems to be related to opposed trends at global and regional scales. The question then arises if this could be somewhat associated with the displacement of the South Atlantic High Pressure reported by several authors (Liu et al., 2007;Reichler, 2009;Bidegain et al., 2014Bidegain et al., , 2015. The very strong correlation of observed wind speed at Montevideo and sea level at Punta del Este (-0.88) seems to be the consequence of coevolution rather than causality. The moderate (+0.43) relationship of total inflow (Q F ) with sea level at Punta del Este shows some degree of influence of hydroclimatic trends and fluctuations on sea level along the Uruguayan coast, particularly for strong events and residues.