Volume 46, Issue 11 p. 5855-5863
Research Letter
Free Access

Fresh Submarine Groundwater Discharge to the Near-Global Coast

YaoQuan Zhou

YaoQuan Zhou

School of Earth Sciences, Ohio State University, Columbus, OH, USA

Search for more papers by this author
Audrey H. Sawyer

Corresponding Author

Audrey H. Sawyer

School of Earth Sciences, Ohio State University, Columbus, OH, USA

Correspondence to: A. H. Sawyer,

[email protected]

Search for more papers by this author
Cédric H. David

Cédric H. David

Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA

Search for more papers by this author
James S. Famiglietti

James S. Famiglietti

Global Institute for Water Security, University of Saskatchewan, Saskatoon, Saskatchewan, Canada

Search for more papers by this author
First published: 24 April 2019
Citations: 68

Abstract

The flow of fresh groundwater to the ocean through the coast (fresh submarine groundwater discharge or fresh SGD) plays an important role in global biogeochemical cycles and coastal water quality. In addition to delivering dissolved elements from land to sea, fresh SGD forms a natural barrier against salinization of coastal aquifers. Here we estimate groundwater discharge rates through the near-global coast (60°N to 60°S) at high resolution using a water budget approach. We find that tropical coasts export more than 56% of all fresh SGD, while midlatitude arid regions export only 10%. Fresh SGD rates from tectonically active margins (coastlines along tectonic plate boundaries) are also significantly greater than passive margins, where most field studies have been focused. Active margins combine rapid uplift and weathering with high rates of fresh SGD and may therefore host exceptionally large groundwater-borne solute fluxes to the coast.

Key Points

  • Almost half of fresh submarine groundwater discharge (SGD) enters the ocean at wet equatorial regions
  • High-relief, tectonically active margins have higher ratios of fresh SGD to river discharge
  • Total annual volume of fresh SGD is ~489 km3/year, or ~1% of river discharge

Plain Language Summary

Fresh groundwater flows from land to sea through coastal rocks and sediment. While the amount of groundwater flow is small compared with rivers, it plays an important role in carrying dissolved chemicals like nutrients to sea, and it helps protect aquifers against salinization. We estimate groundwater flow through the near-global coast. Tropical regions have more groundwater flow, while dry midlatitudes have less. Dry midlatitude regions are therefore more vulnerable to salinization, which is problematic because these regions are also more likely to depend on groundwater to meet their water needs. Additionally, mountainous coastlines along tectonic plate boundaries have relatively large rates of groundwater discharge and may be associated with higher dissolved chemical fluxes to the coast.

1 Introduction

Rivers are the arteries of continents and are responsible for 90–99% of all fresh water that discharges to coasts (Church, 1996). The remaining 1–10% discharges directly from aquifers to coastal wetlands, beaches, and continental shelves (Burnett et al., 2003) in a process known as fresh submarine groundwater discharge (fresh SGD). In addition to fresh SGD that originates onshore, saline groundwater also circulates through the seabed at high rates. In total, the global rate of fresh and saline SGD may be as large as 300–400% of river discharge (Kwon et al., 2014). Radioisotope techniques such as radium isotopes (Moore, 2003) have made it possible to map the distributions of predominantly salty SGD at an unprecedented scale, while distributions of fresh (land-derived) SGD remain elusive. Direct measurements with seepage meters are time intensive and tend to vary over small spatial scales, posing a challenge for upscaling fresh SGD rates.

Fresh SGD is important because it can contain high concentrations of land-derived solutes. For example, fresh SGD from young volcanic rocks may deliver as much as 30% of the global silicate load from land to oceans (Rad et al., 2007). Fresh SGD also introduces nutrients to the coast from onshore fertilizer application. In some areas, nitrate loading from SGD exceeds both atmospheric and riverine sources (Swarzenski et al., 2001). An improved understanding of SGD rates can therefore help refine global biogeochemical budgets and manage coastal water resources.

Fresh SGD also buffers coastal aquifers against salinization. Fresh groundwater floats above denser saline groundwater, and the position of the freshwater-saltwater interface lies in a delicate balance (Michael et al., 2017). The flow of fresh groundwater toward the sea resists landward transport of salt and maintains a steep salinity front. When coastal communities extract groundwater, a portion of the extracted groundwater deducts from fresh SGD. If groundwater extraction reduces fresh SGD below a critical threshold, salinization occurs (Mazi et al., 2013). Aquifers with greater rates of recharge and fresh SGD can generally sustain more groundwater pumping. Thus, an improved understanding of coastal recharge and fresh SGD rates can help manage coastal aquifers.

Computer models have the potential to transform our understanding of fresh SGD, as the availability and quality of hydrologic and topographic data have improved. Fresh SGD rates were recently estimated at subkilometer resolution over data-rich regions using a water budget approach (Destouni et al., 2008; Sawyer et al., 2016; Zhou et al., 2018), but similar global estimates are lacking. We here estimate the flux of fresh SGD to the near-global coast at an unprecedented spatial resolution of 15 arc sec (approximately 500 m at the equator) and show that most fresh SGD is focused in equatorial regions. Additionally, the average rate of fresh SGD along tectonic plate boundaries (active margins) is approximately double the average rate along coastlines far from plate boundaries (passive margins).

2 Methods

We estimate fresh SGD using the water budget approach of Sawyer et al. (2016) and Zhou et al. (2018). Briefly, we define the aquifer control volume as the contributing area, or recharge zone, for fresh SGD. If patterns of groundwater flow are similar to overland flow, the recharge zones that contribute groundwater to the coast are the same as the wedge-shaped coastal catchments that lack streams and instead contribute runoff directly to the coast (Figure 1). This assumption is best for unconfined, thick, homogeneous aquifers in wet climates (Haitjema & Mitchell-Bruker, 2005) and may underestimate fresh SGD in areas with complex geology, especially in dry regions (Sawyer et al., 2016; Zhou et al., 2018). Assuming modest groundwater injection or withdrawal, the annual volume of fresh SGD (QSGD) for each coastal catchment is then the linear average annual net recharge rate (r), or recharge adjusted for evapotranspiration losses, integrated across the recharge area (A; m2):
urn:x-wiley:00948276:media:grl58901:grl58901-math-0001(1)
Details are in the caption following the image
Example of fresh SGD rates and coastal catchment hydrography over tectonically (a) active and (b) passive margins with similar net recharge rates (approximately 0.16 m/year). SGD = submarine groundwater discharge.
The fresh SGD flux (qSGD) can also be defined as
urn:x-wiley:00948276:media:grl58901:grl58901-math-0002(2)
where L is the coastline length for the catchment.

To solve the water budget for coastal aquifers, we approximate the net recharge rate adjusted for evaporative losses (r in equation 1) as the average infiltrating runoff from three land surface models (MOSAIC, NOAH, and VIC) obtained from NASA's Global Land Data Assimilation System (Rodell et al., 2004). All three land surface models solve a surface water budget to estimate the Earth's terrestrial water cycling. None of these models explicitly solves for lateral groundwater flow. Rather, they employ one-dimensional vertical water budgets and hence lack horizontal subsurface water transfers among modeled grid cells. Despite this simplification, the areal base flow contribution in these models is conceptually similar to net recharge (the recharge rate adjusted for evaporative loss), since it represents the flux of water from the soil compartment to deeper storage, which becomes discharge to drainage features. The development of land surface models started half a century ago (Manabe, 1969), and these models continue to benefit from new improvements and undergo detailed calibration and validation (Rodell et al., 2004; Xia, Mitchell, Ek, Cosgrove, et al., 2012; Xia, Mitchell, Ek, Sheffield, et al., 2012) to guarantee appropriate partitioning and closure of the terrestrial water budget.

Coastal recharge areas are delineated based on HydroSHEDS (Hydrological data and maps based on Shuttle Elevation Derivatives at multiple Scales), a high-resolution, near-global map of rivers and their catchments with coverage between 60°N and 60°S. The development of HydroSHEDS is primarily based on elevation data obtained from the NASA Shuttle Radar Topography Mission (Farr et al., 2007). Rivers that have upstream contributing areas greater than 8 km2 are matched with the Global Lake and Wetland Database in ArcWorld to align with river networks. Smaller rivers are delineated solely from the elevation surface based on a flow direction and accumulation method (Tarboton, 1997) that involves a threshold in upstream number of grid cells. Because global topographic data sets are referenced to a spheroidal Earth, a difference in grid cell sizes at varying latitudes generates a discrepancy in the area (though not the number of grid cells) of the smallest catchments that are home to a river reach. Such discrepancies have a potential to generate an underestimation of our fresh SGD calculation at high latitudes. However, in the absence of observed hydrographic data sets—which are only nascent at the continental scale (Allen & Pavelsky, 2015) and expected from future satellite missions (Alsdorf et al., 2007)—the correction of such discrepancy will remain a challenge.

Two components are needed from HydroSHEDS: the coastal catchment area and the coastline length. We first merge all catchments at the continental scale to define the coastline. Catchments in HydroSHEDS that do not contain streams are defined as “coastal” and extracted to yield coastal catchments (Figure 1). Fresh SGD is computed by multiplying catchment area by the annual net recharge rate at the catchment centroid (equation 1). Fresh SGD estimates are sensitive to the choice of hydrographic data set (Destouni et al., 2008; Zhou et al., 2018), but HydroSHEDS is currently the only map product for delineating coastal catchments with near-global coverage. For comparison, the estimated average flux of fresh SGD for the United States Atlantic and Gulf Coast is 2.4 times less using HydroSHEDS than NHDPlus, another hydrographic data set that is only available within the United States (Zhou et al., 2018).

Wherever single values of QSGD or qSGD are reported, they represent ensemble averages for the three recharge models (MOSAIC, NOAH, and VIC). Reported error is calculated as the standard deviation of estimates from MOSAIC, NOAH, and VIC, similar to the approach of Famiglietti et al. (2011) and Reager et al. (2016). This error only accounts for the uncertainty in net recharge, which directly influences fresh SGD (equation 1). The challenge of accurately estimating recharge has been noted by Scanlon et al. (2002). Estimates from the individual recharge models generally agree to within a factor of 2 to 4, and median fresh SGD rates for the near-global coast are 72, 161, and 37 m2/year for MOSAIC, NOAH, and VIC, respectively. These discrepancies may seem large, but they are reasonable, given that (1) fresh SGD rates vary by orders of magnitude and are lognormally distributed (Sawyer et al., 2016), and (2) SGD measurements from multiple methods at the same site can show similar or even larger discrepancies (Burnett et al., 2006).

Simplifying assumptions in the water budget calculation also contribute to uncertainties. Our estimates of fresh SGD are likely low under several conditions: (1) in layered unconfined-confined aquifer systems, where a fraction of fresh SGD originates from deeper confined aquifers with more distal recharge zones; (2) in dry coastal regions which tend to receive groundwater import from upland basins (Schaller & Fan, 2009); (3) in karst, which represents 10% of the world's aquifers (Chen et al., 2017) and has complex patterns of recharge and conduit flow. Our estimates are likely high under other conditions: (1) in areas where groundwater discharges to unmapped tidal creeks or nearshore wetlands (this discharge is generally not considered “submarine” (Taniguchi et al., 2002), but is included with fresh SGD in the water budget method) and (2) in areas of intense groundwater extraction, which deducts from fresh SGD but can also enhance recharge. The combined error associated with our assumptions is difficult to quantify for the near-global coast. However, in a comparison of 10 sites from the continental United States (Bokuniewicz, 1980; Bokuniewicz et al., 2004; Hays & Ullman, 2007; Mulligan & Charette, 2006; Reay et al., 1992; Russoniello et al., 2013; Santos et al., 2009; Simmons, 1992; Uddameri et al., 2014; Zimmermann et al., 1985), our water budget analysis yields similar estimates of fresh SGD as seepage meter studies (Sawyer et al., 2016), other water budget calculations, and validated three-dimensional groundwater flow models (Befus et al., 2017; Zhou et al., 2018; supporting information Figure S1). Our estimates tend to be lower than field-based estimates, likely because of our conservative approach for delineating coastal recharge areas, which would tend to exclude groundwater imports from upland basins.

To assess regions where our estimates may be unrealistically high, we used Darcy's law to calculate the minimum head gradient that would be required to support the estimated fresh SGD rate in every coastal catchment and compared it with the available topographic gradient. We assumed a maximum bound on transmissivity of 1,000 m2/day (hydraulic conductivity of 10 m/day and aquifer thickness of 100 m), which is below the maximum bound reported for coastal Bangladesh (Harvey, 2002), where our fresh SGD rates could easily be overestimated due to the intensity of groundwater extraction for irrigation. We found that the minimum required head gradient only exceeded the topographic gradient in 0.3% of coastal catchments (including parts of Bangladesh). Given these calculations and our comparisons with local field and modeling studies (supporting information Figure S1), we suggest that our estimated fluxes are within reason, and water budgets represent a practical, computationally efficient approach to estimating fresh SGD over large regions, especially in light of the paucity of hydraulic head data and information on aquifer properties to validate global groundwater models.

3 Results and Discussion

Integrated over the near-global coastline, the total annual volume of fresh SGD is 489 km3/year ±337 km3/year, or 1.3% of river discharge (Dai & Trenberth, 2002), in line with previous estimates (Church, 1996; Zekster & Loaiciga, 1993). This estimate does not include the northern coastline of North America and Eurasia, which could contribute additional fresh SGD, particularly under a warming climate with thawing permafrost. Rates of fresh SGD are highly variable around the world (Figure 2). Averaged by continent, Asia has the largest average rate of fresh SGD, while Europe and Australia have the smallest (Table 1). Rates of fresh SGD are also low in dry regions of Central America, and Northern Africa (Figure 2). Most of the fresh SGD enters the ocean at wet equatorial and temperate regions (Figure 2, left). More than half the near-global flux of fresh SGD occurs within the tropics, and 41% occurs within 10° of the equator. The arid midlatitudes (23–40°) contribute only 10% of the world's fresh SGD (Figure 2, left). On average, the rate of fresh SGD per unit length of coast is 5 times less at arid midlatitudes than near the equator (Figure 3).

Details are in the caption following the image
Map of fresh SGD rates along the near-global coastline. Uncertainty of fresh SGD is estimated to be 177.0 m2/year. (left) Comparison of fresh SGD, blue (this study) and river discharge, red (Dai & Trenberth, 2002) by latitude. Note the different ranges of horizontal axes for river discharge and fresh SGD. SGD = submarine groundwater discharge.
Table 1. Fresh SGD Rate Averaged Over Each Continent
Continent SGD (m2/year) Coastline Length in 103km (percentage of total coast included)
VIC NOAH MOSAIC Ensemble average
Africa 41.9 267.3 98.6 135.9 68.1 (100%)
Asia 124.7 874.8 559.5 519.4 327.4 (79%)
Australia 47.2 229.2 102.0 126.1 91.2 (100%)
Europe 58.4 176.3 99.5 111.4 170.4 (70%)
North America 78.9 332.6 210.6 207.4 255.1 (80%)
South America 79.3 411.8 258.1 249.7 139.0 (100%)
  • Note. Uncertainty of fresh SGD is estimated to be 177.0 m2/year. Due to limitations in data coverage at high latitudes, our analysis does not span the entire coastlines of Asia, Europe, and North America. We have approximated the percentage of analyzed coast for these three continents next to analyzed coastline length. SGD = submarine groundwater discharge.
Details are in the caption following the image
Climate, soil, and topography control fresh SGD trends across latitudes for active, passive, and global (all) margins. Available water is the sum of rainfall and snowmelt minus evapotranspiration and represents how wet or dry the climate is. Infiltration capacity is the fraction of available water that infiltrates and contributes to fresh SGD. Drainage (DL) is coastal catchment area divided by coastline length. Dry midlatitudes are highlighted with red marks. SGD = submarine groundwater discharge.

Near-global trends in fresh SGD largely coincide with climate, but the infiltration capacity of the land surface and coastal morphology also play a role (Figure 3). Soil properties and land use dictate the proportion of available water that can infiltrate and be transmitted as groundwater flow to the coast (Taylor et al., 2013; Figure 3). Average infiltration capacity (calculated as the ratio of infiltrating runoff to available water) is generally equal to or greater than 0.6 over most of the globe. This value is greater than infiltration capacities in many inland regions, likely due to the hydrogeologic properties of high-energy modern marine deposits along coastlines. The infiltration capacity declines at high temperate latitudes (45–60°), where 70% of the land is permafrost (Zhang et al., 1999). This decline reduces the average flux of fresh SGD to the ocean at high temperate latitudes (Figure 3).

Another factor that reduces the SGD flux at high latitudes is the shape of coastal catchments. Long, narrow valleys focus groundwater toward small segments of the coast, often at the heads of embayments. Conversely, peninsulas or protrusions distribute groundwater to a long coastal segment (Figure 1). The drainage length, or average distance from any location in the coastal catchment to the shoreline, is generally smaller at high latitudes. The northern high latitudes are dominated by low-relief coastal plains along the Arctic coast (Bokuniewicz et al., 2003) and small islands along the Pacific coast that disperse groundwater. The southern high latitudes are dominated by numerous, small islands of Tierra del Fuego that also disperse groundwater. The average rate of fresh SGD is therefore greatest near the equator where drainage length, infiltration capacity, and available water are all maximized (Figure 3).

Within a given latitude, fresh SGD rates also tend to be greater along tectonically active margins (Figure 3). In fact, the near-global average fresh SGD rate in active margins is 2 times greater than passive margins. These differences are particularly driven by differences in available water. At atmospheric convergence zones near −60°, 0°, and 60° latitude, active margins tend to be wetter (Figure 3), likely due to orographic uplift and precipitation along mountainous coastal zones. Surprisingly, drainage lengths are not significantly different between active and passive coastal catchments when averaged by latitude (Figure 3). In general, high-relief active margins would be expected to have coastal catchments with longer drainage lengths separating steep river valleys (Figure 1), but average drainage lengths are only significantly longer near −20° latitude. These catchments almost exclusively drain the Andes Mountains of South America (Figure 2). Regardless of catchment geography, the higher relief of active margins would also tend to allow steeper hydraulic head gradients to develop near the coast, which could facilitate greater rates of fresh SGD compared to regions where topography limits recharge (Michael et al., 2013). It is interesting that our water budget analysis does not depend on an estimation of hydraulic head gradients (or even topographic gradients) but nevertheless predicts greater rates of fresh SGD in active margins.

4 Implications for Water Management

Gaps in fresh SGD leave midlatitude coastal regions particularly vulnerable to salinization. These latitudes will suffer most from groundwater salinization because they also lack abundant surface water (Dai & Trenberth, 2002; Syed et al., 2009) and tend to use groundwater during dry years to address water shortages. The dry midlatitudes receive only 17% of global river discharge (Figure 2, left) and are expected to become drier in a changing climate (Held & Soden, 2006). Their coastlines are home to over 800 million people, or 36% of the world's coastal population. As densely populated, dry areas turn increasingly to groundwater, they risk salinization and deterioration of critical water resources. Some of these densely populated regions include cities like Shanghai and Los Angeles, which already face water management challenges. Shanghai has experienced intense water shortage and leans heavily on surrounding provinces for its water supply (Zhao et al., 2016). Los Angeles is projected to have the greatest surface water deficit of all cities by 2050 (Flörke et al., 2018) and uses groundwater to stabilize its water insecurity. Even in wet climates with high recharge rates, densely populated megacities can place a strain on coastal aquifers. For example, Bangkok suffers from both land subsidence and salinization due to extensive groundwater pumping in recent decades (Phien-wej et al., 2006). Other vulnerable areas include small islands, which face substantial water security challenges under rising sea levels (Ferguson & Gleeson, 2012; Michael et al., 2013) and are isolated from external water supplies. Many small islands rely heavily on groundwater for domestic and agricultural uses.

Estimates of fresh SGD also reveal contamination threats to the ocean and the fisheries that coastal populations depend on. The ratio of river discharge to fresh SGD is particularly important (supporting information Figure S2), since rivers and groundwater carry contaminants at different concentration levels. Levels of nitrate in SGD can exceed rivers by orders of magnitude and contribute to harmful algal blooms (Slomp & Van Cappellen, 2004). Mercury contamination of fish habitats has also been linked with groundwater discharge more than river discharge in some regions (Black et al., 2009). The world's largest river deltas have some of the highest rates of river discharge that swamp the typical rates of fresh SGD. Meanwhile, mountainous coastal regions can have rates of fresh SGD that approach river discharge (supporting information Figure S2). Along the southwestern coast of South America, fresh SGD is 19% of river discharge on average. The high relief of this tectonically active margin favors fast groundwater flow toward the Pacific Ocean and truncates the drainage area of rivers that flow to the Pacific Ocean. Other areas with a relative abundance of fresh SGD include tectonically active settings such as the northwestern coast of North America, Philippine Sea, and East China Sea (supporting information Figure S2). Some of these regions are experiencing heavy population growth, urbanization, and agricultural developments. Because groundwater residence times in aquifers span decades, marine waters are vulnerable to delayed changes in chemical inputs from fresh SGD.

5 Conclusions

This study provides the first near-global and spatially distributed high-resolution estimate of fresh groundwater fluxes through the coast and can be used to inform new science on our coastal water resources. The near-global distribution of fresh SGD is highly influenced by climate, with concentrated outflows at wet equatorial and high latitudes and gaps at dry midlatitudes. Large population centers in these dry latitudes are vulnerable to aquifer salinization and must manage groundwater extraction carefully to avoid passing tipping points. Unfortunately, arid population centers are the very regions that depend most heavily on groundwater to meet their resource needs. Fresh SGD is also focused along tectonically active margins. The rapid uplift and weathering rates in these margins may be associated with high subterranean solute fluxes to the coast, but field measurements have focused heavily on passive margins along the Atlantic Ocean to date. More field studies, particularly in dry midlatitudes and wet active margins, are needed to assess the global distribution of fresh SGD rates and chemical fluxes.

Data Availability

Fresh SGD data (shape files for coastal catchments and rates of fresh SGD) have been shared publicly on Zenodo at the website (https://doi.org/10.5281/zenodo.2631971).

Acknowledgments

We thank the Editor and two anonymous reviewers for their recommendations that improved the manuscript. C. H. D. and J. S. F. were supported by the Jet Propulsion Laboratory, California Institute of Technology, under a contract with NASA; including a grant from the SWOT Science Team.