Journal of Environmental Radioactivity 145 (2015) 10e18
Contents lists available at ScienceDirect
Journal of Environmental Radioactivity journal homepage: www.elsevier.com/locate/jenvrad
A GIS typology to locate sites of submarine groundwater discharge John Rapaglia a, *, Carley Grant a, Henry Bokuniewicz b, Tsvi Pick b, Jan Scholten c a
Department of Biology, Sacred Heart University, 5151 Park Avenue, Fairfield, CT 06825, USA School of Marine and Atmospheric Sciences, Stony Brook University, Stony Brook, NY 11794, USA c Institute of Geosciences, Christian Albrechts University of Kiel, Otto-Hahn Platz 1, 24098 Kiel, Germany b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 21 November 2014 Received in revised form 11 March 2015 Accepted 11 March 2015 Available online
Although many researchers agree on the importance of submarine groundwater discharge (SGD), it remains difficult to locate and quantify this process. A groundwater typology was developed based on local digital elevation models and compared to concurrent radon mapping indicative of SGD in the Niantic River, CT USA. Areas of high radon activity were located near areas of high flow accumulation lending evidence to the utility of this approach to locate SGD. The benefits of this approach are threefold: fresh terrestrial SGD may be quickly located through widely-available digital elevation models at little or no cost to the investigator; fresh SGD may also be quantified through the GIS approach by multiplying pixelated flow accumulation with the expected annual recharge; and, as these data necessarily quantify only fresh SGD, a comparison of these data with SGD as calculated by Rn activity may allow for the separation of the fresh and circulated fractions of SGD. This exercise was completed for the Niantic River where SGD as calculated by the GIS model is 1.2 m3/s, SGD as calculated by Rn activity is 0.73e5.5 m3/s, and SGD as calculated via a theoretical approach is 1.8e4.3 m3/s. Therefore fresh, terrestrial SGD accounts for 22e100% of total SGD in the Niantic River. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Submarine groundwater discharge Radon mapping Digital elevation models Geographic information systems
1. Introduction Rivers and terrestrial runoff are generally well-constrained sources of nutrients and pollution into coastal areas. However, the role of groundwater discharge in coastal processes has often been overlooked (Burnett et al., 2006). Submarine groundwater discharge (SGD), both terrestrially derived freshwater and circulated seawater, is important to coastal embayments both chemically and volumetrically (Moore, 2010). SGD is a major pathway for land-derived solutes, and can thus largely influence the ecology of coastal communities and various geochemical cycles (Burnett and Dulaiova, 2003; Stieglitz, 2005). As the concentration of dissolved inorganic nutrients is often elevated in groundwater as compared to surface water, SGD, even in small volumes, can bring a disproportionally large flux of nutrients into a system (Stieglitz, 2005; Dulaiova et al., 2005). SGD, therefore, has been implicated as the cause of many adverse coastal ~ udophenomena in recent years (e.g. Matson, 1993; Gobler and San
* Corresponding author. Tel.: þ1 203 396 8347. E-mail addresses: [email protected] (J. Rapaglia), carleymeryl@gmail. com (C. Grant), [email protected] (H. Bokuniewicz), tpick@ stonybrook.edu (T. Pick), [email protected] (J. Scholten). http://dx.doi.org/10.1016/j.jenvrad.2015.03.016 0265-931X/© 2015 Elsevier Ltd. All rights reserved.
Wilhelmy, 2001; Slomp and Van Cappellen, 2004; Laurier et al., 2007; Lee and Kim, 2007). High nutrient, organic carbon, metal and other pollutant loading into coastal lagoons, embayments, estuaries, and open water areas, through SGD is often cited as an important process in the degradation of marine and aquatic environments (e.g. Johannes, 1980; Simpson et al., 2004; Swarzenski et al., 2007; Santos et al., 2008; Spruill and Bratton, 2008). In some cases, eutrophication and algal blooms have been connected to SGD due to the influence it has over N, P, and Si concentrations and the ratio among the nutrients (Stieglitz, 2005). SGD may be the main source of nutrients supporting coastal primary productivity (Johannes, 1980). Meanwhile, high amounts of nitrogen in relation to phosphorous in coastal areas could affect nutrient cycling by driving nitrogen limited systems to becoming phosphorous limited, thus increasing the potential of harmful algal growth (Slomp and Van Cappellen, 2004). Furthermore, anthropogenic land use has a great influence on shallow unconfined aquifers. Residential and agricultural development can increase intrusion of synthetic nutrients and other pollutants into continental groundwater (Slomp and Van Cappellen, 2004), thus serving as a conduit for coastal pollution (Burnett and Dulaiova, 2003; Oliveira et al., 2003; Stieglitz, 2005). Although both fresh groundwater and circulated saline groundwater are significant contributors to coastal systems,
J. Rapaglia et al. / Journal of Environmental Radioactivity 145 (2015) 10e18
nutrient enrichment and pollution is largely associated with the terrestrially derived freshwater fraction. Although, the full extent of ecological impacts depends on the total SGD into a basin, the nutrient inputs likely rely more heavily on the fresh groundwater component. It is thus reasonable to suggest that fresh SGD is the more important contributor of the two fractions in terms of influence on coastal communities. This freshwater fraction is carried as “underflow”, that is, the flow of fresh groundwater under the shoreline. The specific distribution of SGD along a shoreline may also be important if the intent is ultimately to manage contaminant inputs, perhaps with permeable reactive barriers (e.g. Robertson and Cherry, 1995) or other remediation technology. However, this distribution is difficult to locate and the use of conventional techniques to measure SGD are either labor intensive and cover small areas (seepage meters and bulk ground conductivity) or relatively expensive to use and marked with uncertainty (geochemical tracers). Herein we discuss a GIS-based approach, which has little to no cost associated with it, is essentially available everywhere, and can help investigators decide where to look for fresh SGD into coastal areas using more conventional techniques. 1.1. Locating and quantifying SGD Hydrological models have been used with success to determine long-term, large scale SGD (Zekster et al., 2007), but most of these models do not take into account circulated seawater and often produce SGD numbers much lower than SGD measured in the field using seepage meters, tracers, or other methods. Geochemical tracers are widely used to quantify SGD as they are able to integrate the measurement over space and time. While several tracers have been considered, radium and radon remain the tracers of choice as they are highly elevated in groundwater compared to surface water, are considered conservative in surface water, have become easy-to-measure, and can be used to locate “hotspots” of SGD (e.g. Stieglitz et al., 2010). Both radium and radon suffer from the inability to collect direct SGD samples for physical and chemical characteristics, and have difficulty in the separation of fresh and circulated SGD. It is suggested that these geochemical tracers should be used in conjunction with seepage meters when possible (Burnett et al., 2006). When this is not possible, or practical, porewater samples from within the subterranean estuary and directly below the sedimentesea interface are collected from wells, push-point samplers, porewater probes, etc. and multiplied by the SGD measurement to determine SGD constituent flux (Santos et al., 2008). Other methods have been considered (and have been, in some cases, successful) in the location and quantification of SGD. Bulk ground conductivity (BGC)dor the ability of sediments and porewater to conduct electricitydhas been utilized in various locations with a good deal of success in locating areas of fresh SGD (Stieglitz et al., 2008). Indeed, when used in conjunction with seepage meter measurements, BGC can be used to quantify SGD over a larger spatial area than seepage meter measurements alone. This approach, however, assumes that only fresh SGD is important. In addition, some attention has been given to the utility of remote sensing (mainly thermal IR imagery) to locate SGD (Lewandowski et al., 2013; Wilson and Rocha, 2012). In theory, groundwater remains at a constant temperature, while surface water temperatures fluctuate. Therefore, in summer (or winter), there is often a detectable difference in groundwater temperature as compared to surface water temperature. Remote sensing however, can be hampered by the inability to quantify discharge, and the fact that it can only be used during certain seasons and in areas where there is a significant flux of terrestrial groundwater into shallow coastal areas.
11
1.2. SGD typology and tidal influence Regardless of the method employed, the measurement of SGD is labor intensive. Direct measurements are usually available at only a few places along the shore (Taniguchi et al., 2002). In order to translate point measurements into budgets for coastal water bodies, measurements at one, or a few places, need to be extrapolated over wider areas. Construction of an SGD typology is one strategy for extrapolating sparse measurements based on distributions of reasonable, surrogate, known parameters (Bokuniewicz, 2001; Bokuniewicz et al., 2003; Buddemeier et al., 2008; Dahl et al., 2007; Dulaiova et al., 2006). A typology selects relevant, well-distributed parameters that can be mapped throughout the study area, and could reasonably be expected to influence the magnitude of SGD (Bokuniewicz, 2001; Bokuniewicz et al., 2003; Buddemeier et al., 2008; Lucena-Moya et al., 2009). As a result, the typology is not a model, but rather an empirical scaling mechanism. Places where coastal recharge is higher or where the hydraulic gradient at the shoreline is steeper for example, might be expected to have higher SGD than other sites. While such parameters, and others, are not themselves measures of SGD, the conceptual model of SGD processes make them reasonable relevant quantities. Groundwater typology has been applied to extrapolate data on a global scale (Bokuniewicz, 2001; Bokuniewicz et al., 2003; Dahl et al., 2007; Rapaglia et al., 2010). Small-scale, local typologies, like the one used here, however allow the effectiveness of the strategy to be better assessed, and potentially tested in the field. Point measurements might be extrapolated with the aid of the typology, or, alternatively, the typology might be used to guide future sampling, or to identify likely “hotspots” along the shoreline that could be SGD locations. The goal of this study is to build a GISbased groundwater typology, based on groundwater underflow in order to compute SGD trends along shorelines. This hypothesis was tested in the Niantic River Estuary, CT. The typology uses the GIS analysis of digital elevation models (DEM), to locate possible areas of SGD coincident with continuous Rn mapping to test the GISbased typology and to also quantify total SGD into the system. The typographic analysis will be compared with Rn measurements to try and separate the fresh and circulated fractions of SGD. Here we will use as a basic parameterization based on calculations of the underflow, that is, the volume flow rate of fresh groundwater from the terrestrial aquifer under the shoreline. The concept is based on modeling exercises (Destouni and Prieto, 2003; Prieto and Destouni, 2005) that have shown that SGD is related to the underflow modified by the tidal range (Fig. 1). Non-tidal SGD seemed to be well represented to groundwater underflow. This relationship seems “quite independent of site-specific details of hydrogeology…” (Destouni and Prieto, 2003). At low values of Qn, there is a slight non-linear relationship (Destouni and Prieto, 2003), but for QN greater than 800 m3/year/m the relationship is fairly linear in the absence of tides (Destouni and Prieto, 2003). We represented the modeling results by a linear regression merely for convenience as:
SGD ¼ 1:1*Q N þ 470;
(1)
in units of m3/year/m. Notably, SGD exists even if QN is zero due to seawater inflow (Destouni and Prieto, 2003). Numerical models that only calculate the fresh groundwater flow based on terrestrial hydraulic gradients without circulation of seawater almost always underestimate SGD, usually substantially (Burnett et al., 2006). Tidally driven circulation of seawater through coastal aquifers, however, appears to dominate SGD (Fig. 1). The hydraulic gradients driving SGD vary in the presence of tidal pumping (Nielsen, 1990;
12
J. Rapaglia et al. / Journal of Environmental Radioactivity 145 (2015) 10e18
Fig. 1. Tidal effects on SGD, SGDaver ¼ average SGD with large tidal oscillation; SGDss ¼ non-tidal SGD at steady state. From Prieto and Destouni, 2005.
Michael et al., 2003; Taniguchi, 2002). During high tide, saltwater flow penetrates into the upper layers of permeable undersaturated sand (Urish and McKenna, 2004) causing an upper saline plume in coastal aquifers sediments (Moore and Wilson, 2005; Xin et al., 2010). Modeling of tidally-induced circulation of groundwater showed an elevated circulation during intertidal flow, increasing the exchange between groundwater underflow, SGD, by three orders of magnitude (Rocha, 2000; Xin et al., 2010). While the freshwater component of SGD is solely dependent on flow of terrestrial recharge under the shoreline, total SGD is elevated further in the presence of the tides (Prieto and Destouni, 2005). In this case, total SGD is most sensitive to the tidal range (da Rocha et al., 2009). If Qn is low in the presence of a tide, the total SGD is dominated by circulated seawater. As discussed earlier, the freshwater component of SGD in this case may be small volumetrically but still a significant source of contaminants because of elevated concentrations in the terrestrial, fresh groundwater. For convenience, the modeling results incorporating diurnal tides with a 1.3 m tidal range (Prieto and Destouni, 2005) were represented by us with an empirical linear regression as:
SGDt ¼ 0:6*Q N þ 2000;
(2)
for QN less than about 3000 m3/m/year. Presumably, a semi-diurnal tide would result in twice the tidal component of SGDt. The importance of circulated seawater in coastal processes is still a topic of discussion in the SGD community. Nevertheless, it is useful given the above information, to be able to separate the fresh and circulated fractions of total SGD. The freshwater portion is necessary to complete hydrological budgets and for water management purposes. It would be a mistake to disregard the circulated portion of SGD as there are many processes which occur in the fresh-salt mixing zone which affect the coastal ecosystem as well. In this paper, we will estimate the freshwater component of SGD from QN and total SGD from geochemical measurements. These will be compared to the empirical predictions. .
Increasing availability of high resolution DEM's allow for the analysis of an area's groundwater accumulation based on topography. The combination of flow directions and magnitude in the vector analysis make it possible to predict underflow hotspots. While the freshwater component of SGD is important in itself, the freshwater underflow is directly related to total SGD. We will present a GIS-based typology based on a DEM and test the typology by a comparison of areas of high flow accumulation with areas of high Rn activity. Radon-222 (222Rn, T½ ¼ 3.83 days) is an inert gas and is produced by radioactive decay of Ra-226 (226Ra, T½ ¼ 1600 years) which is ubiquitous in geological rocks and soils. In seawater 222Rn is very low due to little 226Ra concentrations (~1 Bq/ m3) whereas 222Rn concentrations in groundwater are 1000-times as high. Thus any 222Rn concentration anomaly in near coastal waters is a strong indication of SGD. A quantification of SGD using 222 Rn requires a mass balance in which all 222Rn sources (input via SGD, rivers and sediments) and 222Rn sinks (inventory in coastal waters, loss to the open sea and to the atmosphere) are taken into account (Burnett and Dulaiova, 2003). In this mass balance the parameter the most difficult to determine is the 222Rn input via SGD as most of the 222Rn is added to the groundwater in the mixing zone between freshwater and seawater, and this mixing and resulting 222 Rn sources may be highly variable (Dulaiova et al., 2008). Theoretically, if SGD is driven by the flow of fresh terrestrial water towards the sea, then areas of high Rn, should be adjacent to areas of high terrestrial flow accumulation, assuming that this flow accumulation is not surface water flow. We hypothesize, therefore, that GIS flow accumulation analysis from high resolution DEM's can be used to locate areas of high SGD. In addition, given that the GIS approach necessarily quantifies only freshwater flow, while Rn activity is a proxy for total SGD, utilizing both GIS and Rn may be useful in separating the fresh and circulated fractions of SGD. In order to test these hypotheses, a simultaneous study of Rn activity and GIS analysis of a DEM was performed in a second coastal embayment, the Niantic River, CT.
1.3. GIS application
2. Study site: Niantic River, CT USA
It is commonly observed that the water table generally conforms to the local typography (Toth, 1963). Hence, it is reasonable to use elevation to predict groundwater flow, particularly in areas with unconfined, isotropic aquifers. Groundwater “accumulation” is the convergence of flow lines in the direction of flow. Groundwater accumulation, for example has been shown to occur at convex shorelines in lakes (Cherkauer and McKereghan, 1991; Lewandowski et al., 2013; Winter and Pfannkuch, 1984)
The Niantic River (41.3243 ; 72.1820 ) is located along the southeastern coast of Connecticut, on the eastern end of Long Island Sound (Fig. 2). The Niantic River is a wide, generally shallow (average depth 2.4 m) estuary. The mean tidal range is 0.8 m. The river is 1 km wide at its widest point and 5.5 km long, although the total length of the contorted shoreline is 22.4 km. Measurable salinity can be found throughout the entire system at high tide and in periods of low precipitation. Its western coast is marked by high
J. Rapaglia et al. / Journal of Environmental Radioactivity 145 (2015) 10e18
(~80 m) wooded relief, with little human development. Its eastern coast is marked by gentle rolling (~20 m) terrain and suburban homes. The land is characterized as glacial till and the coastal aquifer is an unconsolidated loose sand and gravel aquifer. The bedrock underlaying the aquifer is pretty uniform on both sides of the Niantic River with medium grained gneiss or schist surrounded by granite as you move further away from the river (Rodgers, 1985). The river can be separated into two sections, the main basin, which is essentially a wide open bay, and the northern neck which is 2.5 km long and about 0.25 km wide. The Niantic River has a small (ca. 50 m wide) inlet to the open Long Island Sound. Average precipitation has been estimated to be 117 cm/year (Nichols, 1913). Average annual groundwater recharge at predevelopment levels for soils of hydrologic group A in coastal Connecticut is estimated to be between 46 and 51 cm (Mullaney, 2004; CT DEEP, 2004). The average annual surface water discharge into Niantic River is 1.12 m3/s (Mullaney, 2013). The day of sampling was cold and rainy with an average air temperature of 7.8 C and wind speeds of 4.9 m/s. 3. Methods
13
Rn levels while steaming along the coastline at 2 to 3 knots. It was essential to maintain a slow speed, as it would ensure higher spatial resolution of the data (Dulaiova et al., 2005). A Trimble Geoexplorer 6000XH was used to collect GPS coordinates with sub-meter accuracy every 6 s. As continental groundwater discharge has been shown to occur closer to shore (Bokuniewicz and Pavlik, 1990; Johannes, 1980; Oliveira et al., 2003), samples were collected throughout the river in order to better understand the entire Rn activity in the river (Fig. 3). Additionally to further increasing the likelihood that the radon measured would represent the terrestrially derived freshwater faction of SGD, measurements were initiated close to high tide. During high tide, hydrostatic pressure increases, and surface water is more likely to infiltrate the sediments (Burnett and Dulaiova, 2003, 2006). Thus, the water that would be seeping out is more likely to be the continental freshwater portion of SGD, subsequently minimizing the Rn from circulated seawater. Water was continuously pumped through the RAD-Aqua system at a rate of 5e7 L/min by a Flojet water system pump from approximately 1 m below the surface. Once Rn levels in the RAD-7 chamber reached equilibrium (after 15 min of pumping), Rn activity was measured by the RAD-7 using a sniff test every 5 min (approximately every 300 m).
3.1. Radon 3.2. ArcGIS10 Radon was continuously measured in the Niantic using an in-air radon monitor (Durridge Company, Inc. RAD-7), modified with a RAD Aqua, an airewater exchanger, in a closed loop system to detect
Radon activity readings were assigned GPS coordinates using time. The GPS coordinates 5 min prior to each radon activity
Fig. 2. Site location, Niantic River.
14
J. Rapaglia et al. / Journal of Environmental Radioactivity 145 (2015) 10e18
activity, with the highest values (>400 Bq/m3) located along the northwest shore of the northern section of the Niantic River, medium values (200e400 Bq/m3) found along the southern shore of the northern section, the northeast fork, and the small southwest basin. All other locations had relatively low values of Rn activity (
Contents lists available at ScienceDirect
Journal of Environmental Radioactivity journal homepage: www.elsevier.com/locate/jenvrad
A GIS typology to locate sites of submarine groundwater discharge John Rapaglia a, *, Carley Grant a, Henry Bokuniewicz b, Tsvi Pick b, Jan Scholten c a
Department of Biology, Sacred Heart University, 5151 Park Avenue, Fairfield, CT 06825, USA School of Marine and Atmospheric Sciences, Stony Brook University, Stony Brook, NY 11794, USA c Institute of Geosciences, Christian Albrechts University of Kiel, Otto-Hahn Platz 1, 24098 Kiel, Germany b
a r t i c l e i n f o
a b s t r a c t
Article history: Received 21 November 2014 Received in revised form 11 March 2015 Accepted 11 March 2015 Available online
Although many researchers agree on the importance of submarine groundwater discharge (SGD), it remains difficult to locate and quantify this process. A groundwater typology was developed based on local digital elevation models and compared to concurrent radon mapping indicative of SGD in the Niantic River, CT USA. Areas of high radon activity were located near areas of high flow accumulation lending evidence to the utility of this approach to locate SGD. The benefits of this approach are threefold: fresh terrestrial SGD may be quickly located through widely-available digital elevation models at little or no cost to the investigator; fresh SGD may also be quantified through the GIS approach by multiplying pixelated flow accumulation with the expected annual recharge; and, as these data necessarily quantify only fresh SGD, a comparison of these data with SGD as calculated by Rn activity may allow for the separation of the fresh and circulated fractions of SGD. This exercise was completed for the Niantic River where SGD as calculated by the GIS model is 1.2 m3/s, SGD as calculated by Rn activity is 0.73e5.5 m3/s, and SGD as calculated via a theoretical approach is 1.8e4.3 m3/s. Therefore fresh, terrestrial SGD accounts for 22e100% of total SGD in the Niantic River. © 2015 Elsevier Ltd. All rights reserved.
Keywords: Submarine groundwater discharge Radon mapping Digital elevation models Geographic information systems
1. Introduction Rivers and terrestrial runoff are generally well-constrained sources of nutrients and pollution into coastal areas. However, the role of groundwater discharge in coastal processes has often been overlooked (Burnett et al., 2006). Submarine groundwater discharge (SGD), both terrestrially derived freshwater and circulated seawater, is important to coastal embayments both chemically and volumetrically (Moore, 2010). SGD is a major pathway for land-derived solutes, and can thus largely influence the ecology of coastal communities and various geochemical cycles (Burnett and Dulaiova, 2003; Stieglitz, 2005). As the concentration of dissolved inorganic nutrients is often elevated in groundwater as compared to surface water, SGD, even in small volumes, can bring a disproportionally large flux of nutrients into a system (Stieglitz, 2005; Dulaiova et al., 2005). SGD, therefore, has been implicated as the cause of many adverse coastal ~ udophenomena in recent years (e.g. Matson, 1993; Gobler and San
* Corresponding author. Tel.: þ1 203 396 8347. E-mail addresses: [email protected] (J. Rapaglia), carleymeryl@gmail. com (C. Grant), [email protected] (H. Bokuniewicz), tpick@ stonybrook.edu (T. Pick), [email protected] (J. Scholten). http://dx.doi.org/10.1016/j.jenvrad.2015.03.016 0265-931X/© 2015 Elsevier Ltd. All rights reserved.
Wilhelmy, 2001; Slomp and Van Cappellen, 2004; Laurier et al., 2007; Lee and Kim, 2007). High nutrient, organic carbon, metal and other pollutant loading into coastal lagoons, embayments, estuaries, and open water areas, through SGD is often cited as an important process in the degradation of marine and aquatic environments (e.g. Johannes, 1980; Simpson et al., 2004; Swarzenski et al., 2007; Santos et al., 2008; Spruill and Bratton, 2008). In some cases, eutrophication and algal blooms have been connected to SGD due to the influence it has over N, P, and Si concentrations and the ratio among the nutrients (Stieglitz, 2005). SGD may be the main source of nutrients supporting coastal primary productivity (Johannes, 1980). Meanwhile, high amounts of nitrogen in relation to phosphorous in coastal areas could affect nutrient cycling by driving nitrogen limited systems to becoming phosphorous limited, thus increasing the potential of harmful algal growth (Slomp and Van Cappellen, 2004). Furthermore, anthropogenic land use has a great influence on shallow unconfined aquifers. Residential and agricultural development can increase intrusion of synthetic nutrients and other pollutants into continental groundwater (Slomp and Van Cappellen, 2004), thus serving as a conduit for coastal pollution (Burnett and Dulaiova, 2003; Oliveira et al., 2003; Stieglitz, 2005). Although both fresh groundwater and circulated saline groundwater are significant contributors to coastal systems,
J. Rapaglia et al. / Journal of Environmental Radioactivity 145 (2015) 10e18
nutrient enrichment and pollution is largely associated with the terrestrially derived freshwater fraction. Although, the full extent of ecological impacts depends on the total SGD into a basin, the nutrient inputs likely rely more heavily on the fresh groundwater component. It is thus reasonable to suggest that fresh SGD is the more important contributor of the two fractions in terms of influence on coastal communities. This freshwater fraction is carried as “underflow”, that is, the flow of fresh groundwater under the shoreline. The specific distribution of SGD along a shoreline may also be important if the intent is ultimately to manage contaminant inputs, perhaps with permeable reactive barriers (e.g. Robertson and Cherry, 1995) or other remediation technology. However, this distribution is difficult to locate and the use of conventional techniques to measure SGD are either labor intensive and cover small areas (seepage meters and bulk ground conductivity) or relatively expensive to use and marked with uncertainty (geochemical tracers). Herein we discuss a GIS-based approach, which has little to no cost associated with it, is essentially available everywhere, and can help investigators decide where to look for fresh SGD into coastal areas using more conventional techniques. 1.1. Locating and quantifying SGD Hydrological models have been used with success to determine long-term, large scale SGD (Zekster et al., 2007), but most of these models do not take into account circulated seawater and often produce SGD numbers much lower than SGD measured in the field using seepage meters, tracers, or other methods. Geochemical tracers are widely used to quantify SGD as they are able to integrate the measurement over space and time. While several tracers have been considered, radium and radon remain the tracers of choice as they are highly elevated in groundwater compared to surface water, are considered conservative in surface water, have become easy-to-measure, and can be used to locate “hotspots” of SGD (e.g. Stieglitz et al., 2010). Both radium and radon suffer from the inability to collect direct SGD samples for physical and chemical characteristics, and have difficulty in the separation of fresh and circulated SGD. It is suggested that these geochemical tracers should be used in conjunction with seepage meters when possible (Burnett et al., 2006). When this is not possible, or practical, porewater samples from within the subterranean estuary and directly below the sedimentesea interface are collected from wells, push-point samplers, porewater probes, etc. and multiplied by the SGD measurement to determine SGD constituent flux (Santos et al., 2008). Other methods have been considered (and have been, in some cases, successful) in the location and quantification of SGD. Bulk ground conductivity (BGC)dor the ability of sediments and porewater to conduct electricitydhas been utilized in various locations with a good deal of success in locating areas of fresh SGD (Stieglitz et al., 2008). Indeed, when used in conjunction with seepage meter measurements, BGC can be used to quantify SGD over a larger spatial area than seepage meter measurements alone. This approach, however, assumes that only fresh SGD is important. In addition, some attention has been given to the utility of remote sensing (mainly thermal IR imagery) to locate SGD (Lewandowski et al., 2013; Wilson and Rocha, 2012). In theory, groundwater remains at a constant temperature, while surface water temperatures fluctuate. Therefore, in summer (or winter), there is often a detectable difference in groundwater temperature as compared to surface water temperature. Remote sensing however, can be hampered by the inability to quantify discharge, and the fact that it can only be used during certain seasons and in areas where there is a significant flux of terrestrial groundwater into shallow coastal areas.
11
1.2. SGD typology and tidal influence Regardless of the method employed, the measurement of SGD is labor intensive. Direct measurements are usually available at only a few places along the shore (Taniguchi et al., 2002). In order to translate point measurements into budgets for coastal water bodies, measurements at one, or a few places, need to be extrapolated over wider areas. Construction of an SGD typology is one strategy for extrapolating sparse measurements based on distributions of reasonable, surrogate, known parameters (Bokuniewicz, 2001; Bokuniewicz et al., 2003; Buddemeier et al., 2008; Dahl et al., 2007; Dulaiova et al., 2006). A typology selects relevant, well-distributed parameters that can be mapped throughout the study area, and could reasonably be expected to influence the magnitude of SGD (Bokuniewicz, 2001; Bokuniewicz et al., 2003; Buddemeier et al., 2008; Lucena-Moya et al., 2009). As a result, the typology is not a model, but rather an empirical scaling mechanism. Places where coastal recharge is higher or where the hydraulic gradient at the shoreline is steeper for example, might be expected to have higher SGD than other sites. While such parameters, and others, are not themselves measures of SGD, the conceptual model of SGD processes make them reasonable relevant quantities. Groundwater typology has been applied to extrapolate data on a global scale (Bokuniewicz, 2001; Bokuniewicz et al., 2003; Dahl et al., 2007; Rapaglia et al., 2010). Small-scale, local typologies, like the one used here, however allow the effectiveness of the strategy to be better assessed, and potentially tested in the field. Point measurements might be extrapolated with the aid of the typology, or, alternatively, the typology might be used to guide future sampling, or to identify likely “hotspots” along the shoreline that could be SGD locations. The goal of this study is to build a GISbased groundwater typology, based on groundwater underflow in order to compute SGD trends along shorelines. This hypothesis was tested in the Niantic River Estuary, CT. The typology uses the GIS analysis of digital elevation models (DEM), to locate possible areas of SGD coincident with continuous Rn mapping to test the GISbased typology and to also quantify total SGD into the system. The typographic analysis will be compared with Rn measurements to try and separate the fresh and circulated fractions of SGD. Here we will use as a basic parameterization based on calculations of the underflow, that is, the volume flow rate of fresh groundwater from the terrestrial aquifer under the shoreline. The concept is based on modeling exercises (Destouni and Prieto, 2003; Prieto and Destouni, 2005) that have shown that SGD is related to the underflow modified by the tidal range (Fig. 1). Non-tidal SGD seemed to be well represented to groundwater underflow. This relationship seems “quite independent of site-specific details of hydrogeology…” (Destouni and Prieto, 2003). At low values of Qn, there is a slight non-linear relationship (Destouni and Prieto, 2003), but for QN greater than 800 m3/year/m the relationship is fairly linear in the absence of tides (Destouni and Prieto, 2003). We represented the modeling results by a linear regression merely for convenience as:
SGD ¼ 1:1*Q N þ 470;
(1)
in units of m3/year/m. Notably, SGD exists even if QN is zero due to seawater inflow (Destouni and Prieto, 2003). Numerical models that only calculate the fresh groundwater flow based on terrestrial hydraulic gradients without circulation of seawater almost always underestimate SGD, usually substantially (Burnett et al., 2006). Tidally driven circulation of seawater through coastal aquifers, however, appears to dominate SGD (Fig. 1). The hydraulic gradients driving SGD vary in the presence of tidal pumping (Nielsen, 1990;
12
J. Rapaglia et al. / Journal of Environmental Radioactivity 145 (2015) 10e18
Fig. 1. Tidal effects on SGD, SGDaver ¼ average SGD with large tidal oscillation; SGDss ¼ non-tidal SGD at steady state. From Prieto and Destouni, 2005.
Michael et al., 2003; Taniguchi, 2002). During high tide, saltwater flow penetrates into the upper layers of permeable undersaturated sand (Urish and McKenna, 2004) causing an upper saline plume in coastal aquifers sediments (Moore and Wilson, 2005; Xin et al., 2010). Modeling of tidally-induced circulation of groundwater showed an elevated circulation during intertidal flow, increasing the exchange between groundwater underflow, SGD, by three orders of magnitude (Rocha, 2000; Xin et al., 2010). While the freshwater component of SGD is solely dependent on flow of terrestrial recharge under the shoreline, total SGD is elevated further in the presence of the tides (Prieto and Destouni, 2005). In this case, total SGD is most sensitive to the tidal range (da Rocha et al., 2009). If Qn is low in the presence of a tide, the total SGD is dominated by circulated seawater. As discussed earlier, the freshwater component of SGD in this case may be small volumetrically but still a significant source of contaminants because of elevated concentrations in the terrestrial, fresh groundwater. For convenience, the modeling results incorporating diurnal tides with a 1.3 m tidal range (Prieto and Destouni, 2005) were represented by us with an empirical linear regression as:
SGDt ¼ 0:6*Q N þ 2000;
(2)
for QN less than about 3000 m3/m/year. Presumably, a semi-diurnal tide would result in twice the tidal component of SGDt. The importance of circulated seawater in coastal processes is still a topic of discussion in the SGD community. Nevertheless, it is useful given the above information, to be able to separate the fresh and circulated fractions of total SGD. The freshwater portion is necessary to complete hydrological budgets and for water management purposes. It would be a mistake to disregard the circulated portion of SGD as there are many processes which occur in the fresh-salt mixing zone which affect the coastal ecosystem as well. In this paper, we will estimate the freshwater component of SGD from QN and total SGD from geochemical measurements. These will be compared to the empirical predictions. .
Increasing availability of high resolution DEM's allow for the analysis of an area's groundwater accumulation based on topography. The combination of flow directions and magnitude in the vector analysis make it possible to predict underflow hotspots. While the freshwater component of SGD is important in itself, the freshwater underflow is directly related to total SGD. We will present a GIS-based typology based on a DEM and test the typology by a comparison of areas of high flow accumulation with areas of high Rn activity. Radon-222 (222Rn, T½ ¼ 3.83 days) is an inert gas and is produced by radioactive decay of Ra-226 (226Ra, T½ ¼ 1600 years) which is ubiquitous in geological rocks and soils. In seawater 222Rn is very low due to little 226Ra concentrations (~1 Bq/ m3) whereas 222Rn concentrations in groundwater are 1000-times as high. Thus any 222Rn concentration anomaly in near coastal waters is a strong indication of SGD. A quantification of SGD using 222 Rn requires a mass balance in which all 222Rn sources (input via SGD, rivers and sediments) and 222Rn sinks (inventory in coastal waters, loss to the open sea and to the atmosphere) are taken into account (Burnett and Dulaiova, 2003). In this mass balance the parameter the most difficult to determine is the 222Rn input via SGD as most of the 222Rn is added to the groundwater in the mixing zone between freshwater and seawater, and this mixing and resulting 222 Rn sources may be highly variable (Dulaiova et al., 2008). Theoretically, if SGD is driven by the flow of fresh terrestrial water towards the sea, then areas of high Rn, should be adjacent to areas of high terrestrial flow accumulation, assuming that this flow accumulation is not surface water flow. We hypothesize, therefore, that GIS flow accumulation analysis from high resolution DEM's can be used to locate areas of high SGD. In addition, given that the GIS approach necessarily quantifies only freshwater flow, while Rn activity is a proxy for total SGD, utilizing both GIS and Rn may be useful in separating the fresh and circulated fractions of SGD. In order to test these hypotheses, a simultaneous study of Rn activity and GIS analysis of a DEM was performed in a second coastal embayment, the Niantic River, CT.
1.3. GIS application
2. Study site: Niantic River, CT USA
It is commonly observed that the water table generally conforms to the local typography (Toth, 1963). Hence, it is reasonable to use elevation to predict groundwater flow, particularly in areas with unconfined, isotropic aquifers. Groundwater “accumulation” is the convergence of flow lines in the direction of flow. Groundwater accumulation, for example has been shown to occur at convex shorelines in lakes (Cherkauer and McKereghan, 1991; Lewandowski et al., 2013; Winter and Pfannkuch, 1984)
The Niantic River (41.3243 ; 72.1820 ) is located along the southeastern coast of Connecticut, on the eastern end of Long Island Sound (Fig. 2). The Niantic River is a wide, generally shallow (average depth 2.4 m) estuary. The mean tidal range is 0.8 m. The river is 1 km wide at its widest point and 5.5 km long, although the total length of the contorted shoreline is 22.4 km. Measurable salinity can be found throughout the entire system at high tide and in periods of low precipitation. Its western coast is marked by high
J. Rapaglia et al. / Journal of Environmental Radioactivity 145 (2015) 10e18
(~80 m) wooded relief, with little human development. Its eastern coast is marked by gentle rolling (~20 m) terrain and suburban homes. The land is characterized as glacial till and the coastal aquifer is an unconsolidated loose sand and gravel aquifer. The bedrock underlaying the aquifer is pretty uniform on both sides of the Niantic River with medium grained gneiss or schist surrounded by granite as you move further away from the river (Rodgers, 1985). The river can be separated into two sections, the main basin, which is essentially a wide open bay, and the northern neck which is 2.5 km long and about 0.25 km wide. The Niantic River has a small (ca. 50 m wide) inlet to the open Long Island Sound. Average precipitation has been estimated to be 117 cm/year (Nichols, 1913). Average annual groundwater recharge at predevelopment levels for soils of hydrologic group A in coastal Connecticut is estimated to be between 46 and 51 cm (Mullaney, 2004; CT DEEP, 2004). The average annual surface water discharge into Niantic River is 1.12 m3/s (Mullaney, 2013). The day of sampling was cold and rainy with an average air temperature of 7.8 C and wind speeds of 4.9 m/s. 3. Methods
13
Rn levels while steaming along the coastline at 2 to 3 knots. It was essential to maintain a slow speed, as it would ensure higher spatial resolution of the data (Dulaiova et al., 2005). A Trimble Geoexplorer 6000XH was used to collect GPS coordinates with sub-meter accuracy every 6 s. As continental groundwater discharge has been shown to occur closer to shore (Bokuniewicz and Pavlik, 1990; Johannes, 1980; Oliveira et al., 2003), samples were collected throughout the river in order to better understand the entire Rn activity in the river (Fig. 3). Additionally to further increasing the likelihood that the radon measured would represent the terrestrially derived freshwater faction of SGD, measurements were initiated close to high tide. During high tide, hydrostatic pressure increases, and surface water is more likely to infiltrate the sediments (Burnett and Dulaiova, 2003, 2006). Thus, the water that would be seeping out is more likely to be the continental freshwater portion of SGD, subsequently minimizing the Rn from circulated seawater. Water was continuously pumped through the RAD-Aqua system at a rate of 5e7 L/min by a Flojet water system pump from approximately 1 m below the surface. Once Rn levels in the RAD-7 chamber reached equilibrium (after 15 min of pumping), Rn activity was measured by the RAD-7 using a sniff test every 5 min (approximately every 300 m).
3.1. Radon 3.2. ArcGIS10 Radon was continuously measured in the Niantic using an in-air radon monitor (Durridge Company, Inc. RAD-7), modified with a RAD Aqua, an airewater exchanger, in a closed loop system to detect
Radon activity readings were assigned GPS coordinates using time. The GPS coordinates 5 min prior to each radon activity
Fig. 2. Site location, Niantic River.
14
J. Rapaglia et al. / Journal of Environmental Radioactivity 145 (2015) 10e18
activity, with the highest values (>400 Bq/m3) located along the northwest shore of the northern section of the Niantic River, medium values (200e400 Bq/m3) found along the southern shore of the northern section, the northeast fork, and the small southwest basin. All other locations had relatively low values of Rn activity (
