Offshore fresh groundwater in coastal unconsolidated sediment systems as a potential fresh water source in the 21st century

During recent decades, coastal communities settled along the global coastline have been subject to rising fresh water stress caused by both natural (Oude Essink et al 2010, Faneca Sànchez et al 2012, Rasmussen et al 2013, Yu et al 2016, Yang et al 2018) and anthropogenic (Syvitski et al 2009, Yamanaka et al 2011, Giosan et al 2014, van Camp et al 2014, Tessler et al 2015, Minderhoud et al 2017) causes. It might appear that the onshore and offshore zones of the global coast are geographically clearly defined as sea level stayed relatively constant (Lambeck et al 2014) throughout human recorded history. However, this coastal boundary is in reality rather dynamic when considering a larger geologic time scale. Before around 20 000 years ago, during the Last Glacial Maximum, global mean sea level was more than 120 m lower than current sea level. As a result, extensive continental shelf areas, nowadays submerged under shallow seas, were exposed to terrestrial conditions for tens of thousands of years and hosted an environment full of rivers and fresh water lakes (Head and Gibbard 2005). Under such circumstances, vast fresh groundwater volumes developed through precipitation (i.e. meteoric water), via increased fresh groundwater inflow from inland due to greater groundwater head gradient and through extensive fresh surface water systems covering the current continental shelf areas.

Following the first global overview of offshore fresh groundwater (OFG) (Post et al 2013), which is based on numerous case studies since 1979 (Hathaway et al 1979), several regional scale studies (Levi et al 2018, Morgan et al 2018, Gustafson et al 2019, Thomas et al 2019) have been conducted in recent years. It is important to mention that some other studies (Micallef et al 2020, 2021) define OFG as offshore freshened groundwater that also includes brackish levels of salinity; in our study OFG represents offshore fresh groundwater (<1 gl l⁻¹ TDS (total dissolved solids)). It has been documented that OFG can be found tens or even hundreds of kilometers offshore while reaching depths of up to several kilometers. One of the most thoroughly studied and documented OFG occurrence is in the continental shelf off the New Jersey coast (Meisler et al 1984, Gustafson et al 2019, Thomas et al 2019). An extensive data acquisition of both geological and geophysical data was combined with numerical groundwater flow modelling (Gustafson et al 2019, Thomas et al 2019) to quantify the OFG volume stored in the porous unconsolidated sediments deposited at the continental shelf. OFG can also be stored in fractured karstic rocks as has been documented in e.g. Western and South-eastern Australia (Varma and Michael 2011, Morgan et al 2018). The abovementioned studies show that OFG is often preserved under low permeable geological layers that limit vertical seawater infiltration. These complex heterogeneous geological systems are found in both unconsolidated and karstic systems, stressing the need for a variety of geological information sources to successfully reconstruct regional offshore groundwater conditions. In our study, we combine global thickness estimates (Zamrsky et al 2018) and geological heterogeneity assessments (Zamrsky et al 2020) of unconsolidated sediment systems with variable-density groundwater flow and coupled salt transport modelling to estimate OFG volumes of the global coast (see section 2 and supporting information (SI) figure 1 (available online at stacks.iop.org/ERL/17/014021/mmedia)). Our study is solely focused on unconsolidated sediment systems, due to the lack of globally available data and the high variation in local characteristics of karstic systems (Chen et al 2017) (e.g. their complex conduit networks). Thus, when referring to global OFG volumes we refer to those in unconsolidated coastal aquifers only.

These regional geological scenarios, alongside other hydrogeological parameters (elevation, groundwater recharge estimation), are used for quantifying regional OFG volumes and their uncertainty by means of cross-sectional 2D variable-density groundwater flow and coupled salt transport models (called 'groundwater models' hereafter; SI figure 3). The temporal resolution of our numerical models covers the sea level fluctuation over one full glacial-interglacial cycle (Grant et al 2012) (approximately 125 000 years). This allows us to simulate past fresh water infiltration into the aquifers and aquitards that make up the present-day continental shelves during sea level low stands, and the subsequent salinization of fresh groundwater resources, or their preservation under less permeable clay deposits, during fast rising sea levels in the past 20 000 years (Lambeck et al 2014). Herein, we measure salinity as total dissolved solids (g l⁻¹ TDS) with fresh groundwater threshold defined as 1 g l⁻¹ TDS which is the upper limit considered acceptable to consumers, as defined by World Health Organization (2017). Further details are provided in the section 2 and the SI.

2.1. Regional representative coastal groundwater flow models

2.2. OFG volume estimation

The groundwater salinity profiles of each simulation are aggregated for all ARPs' 24 groundwater models based on the associated 24 geological scenarios, see SI figure 1(C). The mean groundwater salinity concentration value is then calculated for each groundwater model cell resulting in a mean groundwater salinity profile for each time step throughout the whole simulation duration. A threshold of 1 g l⁻¹ TDS is chosen as a boundary between fresh and brackish water which is in line with previous studies (Post et al 2013). The OFG volumes estimated for all ARP groundwater models for the same COSCAT region are aggregated to calculate the total OFG volumes ${V_{{\text{total}}}}$ for the COSCAT region, see equation (1):

$$\begin{equation}{V_{{\text{total}}}} = { }{L_{{\text{off}}}} \times { }{D_{{\text{shelf}}}} \times { }{L_{{\text{coast}}}} \times \frac{{{\text{FG}}{{\text{F}}_{{\text{mean}}}}}}{{100}} \times p\end{equation} \tag{ 1 }$$

where ${L_{{\text{off}}}}$ stands for modelled offshore distance from the coastline (km), ${D_{{\text{shelf}}}}$ represents the average unconsolidated sediment thickness at the coastline (km), ${L_{{\text{coast}}}}$ is the coastal length (km) of each coastal type within the COSCAT region and is then matched with the corresponding ARP's mean fresh groundwater fraction (${\text{FG}}{{\text{F}}_{{\text{mean}}}}$) to translate the cross-sectional estimates into a 3D volume (km³), with p a constant porosity of 0.3 (−). The constant porosity value is based on several previous studies (Groen et al 2000, Oude Essink 2001, van Camp et al 2014, Zamrsky et al 2020). The final OFG volumes per COSCAT region presented in this study are the sum of all ARPs located within the COSCAT region. The shallow OFG volume is estimated as all fresh groundwater located offshore and above 300 meters below sea level (m bsl) (see SI figure 9), chosen arbitrarily as a depth when economically viable groundwater pumping is feasible (i.e. the infrastructure costs are lower than the value of extracted fresh water). Equation (1) is repeated by using ${\text{FG}}{{\text{F}}_{{\text{min}}}}$ and ${\text{FG}}{{\text{F}}_{{\text{max}}}}$ from the 24 groundwater models of each ARP to obtain uncertainty ranges per COSCAT region as well as for the global total OFG volume. The uncertainty of the latter is inflated because it assumes that geological uncertainties between COSCAT regions are fully correlated.

Potentially, brackish offshore groundwater could also be regarded as potential future water resource (Post et al 2013, Alsarayreh et al 2021, Arico et al 2021). That is why, additionally, the volumes of brackish offshore groundwater were also estimated using equation (1) above with OFG fraction representing volume of groundwater with concentrations lower than 10 TDS g l⁻¹. The OFG is then subtracted from the calculated brackish offshore groundwater so only groundwater with concentrations between 1 and 10 TDS g l⁻¹ is taken into account as brackish. These brackish offshore groundwater volumes are shown in SI figure 10.

We also estimated OFG renewal time scales by fresh groundwater inflow from the landwards direction through calculating a coastal fresh groundwater discharge flux by summing up the lateral seaward flux in cells with fresh groundwater concentration at the coastline location. Since our SEAWAT models are 2D representations of the coastal groundwater flow conditions, the resulting coastal fresh groundwater discharge flux values are multiplied by the coastal length and the average unconsolidated sediment thickness at the coastline for each COSCAT region to estimate the total coastal fresh groundwater discharge flux (km³ yr⁻¹) for given coastal region. The average renewal time for a given COSCAT region is then calculated as OFG volume divided by the total coastal fresh groundwater discharge flux based on the definition of Margat and van der Gun (2013). This yields the estimated time it would take to fill the estimated OFG volume under current groundwater flow conditions. Groundwater systems with renewal times larger than 1000 years are considered non-renewable (Margat and van der Gun 2013).

2.3. Regional water demand and onshore groundwater extractions

Summing up all regional OFG volume estimates we approximate the global OFG volume to be 1.06 ± 0.2 × 10⁶ km³ (⩽1 g l⁻¹ TDS). However, since our groundwater models have a constant porosity value (0.3), we assess the porosity value effect on estimated OFG volume by carrying out a groundwater sensitivity study with varying porosity value (0.1, 0.2, 0.4) in a randomly selected region. We can confirm that our groundwater models with the lowest porosity (0.1) show less than half OFG fractions than the case with porosity of 0.3 (see SI figure 4). This suggests that the estimated uncertainty about the global OFG volume may be underestimated due to constant porosity considered. With that in mind, our estimated global OFG volume amounts to about 5%–10% of total fresh groundwater worldwide (estimated to be between 10.5 × 10⁶ km³ (Gleeson et al 2016) and 21.8 × 10⁶ km³ (Ferguson et al 2021)) and roughly 3% of total global fresh water (Shiklomanov 1998). The OFG volumes are most likely non-renewable as they were deposited during glacial periods when sea levels were substantially lower compared to current sea level. Indeed, when we calculate renewal times by the ratio of OFG volume and total coastal fresh groundwater discharge from the groundwater models (SI figure 11), we see that renewal rates exceed 1000 years for most of the coastal regions, rendering the OFG volumes effectively non-renewable.

The estimated OFG volumes vary considerably between the world's coastal regions (figure 1). Regions such as North-east USA, North Australia and South-east Asia accommodate substantial OFG volumes. However, due to unknown heterogeneity in geology and other hydrological factors (e.g. groundwater recharge, influence of rivers), the uncertainties in OFG volumes can be considerable for certain regions, e.g. the coasts around Arabian Peninsula, the Gulf Coast and the North Sea basin (SI figure 5). To put these OFG volume estimates into perspective, we compared them with extracted present day total water demand per year from simulations with a global hydrology and water resources model (Sutanudjaja et al 2018) for a 200 km inland buffer in each coastal region. South and South-east Asia stand out as regions with high current water demand and large OFG volumes, which calls for further investigations into more accurate quantification and potential exploration of these volumes. Water demand is projected to increase steadily during 21st century due to population growth and climate change (Prudhomme et al 2014, Wada and Bierkens 2014), and OFG could prove a vital additional fresh water source in regions such as West Africa and East South America as well (SI figure 6).

Zoom In Zoom Out Reset image size

We compared our OFG estimates with several regional scale studies (Oteri 1988, Groen et al 2000, Person et al 2003, 2012, Varma and Michael 2011, Zhang et al 2011, Bakari et al 2012, Amir et al 2013, Geldern et al 2013, Pauw et al 2014, Jiao et al 2015, Larsen et al 2017, Haroon et al 2018, Paleologos et al 2018, Engelen et al 2019, Gustafson et al 2019, Knight et al 2019, Thomas et al 2019, Bertoni et al 2020) and a set of offshore observation wells (Micallef 2020) limited to areas with unconsolidated sediments. Unfortunately, the majority of these regional scale studies do not provide volume estimates. Therefore, OFG's distance from coastline and depth below sea level were used as proxies. Several studies only provide one of the two proxies, while few only show OFG observations without measurements (see figure 2(A)). Fifteen regional studies covering all continents are taken into account, most of which contain both depth and distance from coast information. Our estimates are in the same order of magnitude as the observed values (figure 2(B)), with only few areas showing large discrepancies. This mostly concerns estimated thickness of OFG where our study shows thinner OFG bodies offshore as compared to literature sources. The offshore observation well dataset (Micallef 2020) gives a point measurement at a certain offshore distance and also the observed top and bottom OFG elevation. The OFG extent is then extracted from the corresponding groundwater model of the same region at the same offshore distance as each observation well (figure 2(C)). While in the same order of magnitude, in several cases we observe a rather large difference (hundreds up to few thousands of meters) in estimated and observed OFG depth (wells ID 16, 20, 22, 23, 24, 27, 34). However, for 80% of the well locations our estimated OFG depth extents overlap the observed OFG depth ranges.

Zoom In Zoom Out Reset image size

Our global OFG volume estimate is in line with a previous global OFG assessment (Post et al 2013), which considers a more generous fresh water threshold (<10 g l⁻¹ TDS) and estimated global OFG volume along passive continental margins to be 0.3 × 10⁶ km³. However, this value may vary up to factor of two (Post et al 2013) due to uncertainty about porosity values ranging between 0.03–0.4, and explains our higher estimate which is based on a constant porosity of 0.3. Recent study by Micallef et al (2021) provides an estimate of offshore freshened groundwater with higher salinity (<10 g l⁻¹ TDS) than considered in our study (<1 g l⁻¹ TDS). Our modelling results show a total volume of groundwater with salinity below 10 g l⁻¹ TDS to be 1.24 ± 0.19 × 10⁶ km³, which is approximately 6 times higher than the estimate by Micallef et al (2021) (0.2 × 10⁶ km³). However, Micallef et al (2021) note that the actual volume of offshore freshened groundwater (and therefore also the OFG volume) is very likely higher than their estimate due to their methodology only considering offshore groundwater emplaced by meteoric recharge and their measurement database often does not provide the full extent of the offshore freshened groundwater bodies. This suggests that our study possibly overestimates the OFG volume along the global coastline but nevertheless provides an estimate in the same order of magnitude as previous global studies (based mostly on qualitative analyses). This can be partly explained by simplifications made during our groundwater modelling study (24 geological scenarios per coastal region) and subsequent OFG volume calculation for each coastal region, where we assume a single cross-sectional groundwater salinity distribution over lengthy coastline stretches. Comparing regional cross-sectional OFG schematizations with our groundwater model results shows a relatively good match in terms of OFG extent and depth (figure 2(B)), but we can also observe some discrepancies presumably due to the large-scale nature of our analysis. As expected, this scale discrepancy is more pronounced when comparing top and bottom of OFG estimates with local observation wells (figure 2(C)). Furthermore, since in our study we only consider unconsolidated sediment systems, older unconsolidated sediment aquifers overlayed by rock formations (e.g. sandstone, limestone) are not accounted for in our approach. This can lead to missing OFG reserves in the deeper formations by our groundwater models as is the case for the Gippsland basin in South Australia (Varma and Michael 2011) (see figure 2(C), observation well 27). Other reasons for larger discrepancies can be caused by the scale difference between local point measurements and our average OFG estimates for regions spanning over hundreds (up to thousands) of kilometers of coastline. It is also possible that the individual observation well measurements were only conducted at a certain depth and do not provide the full extent of the actual range of OFG extent at a given location. Despite this, a satisfactory match is observed in shallower unconsolidated sediment depths (up to 500 m bsl.) which suggests that our approach can be used to estimate regional OFG volumes.

While our offshore fresh (or freshened) groundwater volume estimates are accordance with previous studies, there are a number of limitations and uncertainties that have to be addressed. First, the extent of the cross-sectional profiles is limited to up to 200 km offshore. The majority of the offshore freshened groundwater occurrences have been reported to lie within 55 km distance from the current coastline but the maximum distance has been reported to be 720 km (Micallef et al 2021). This means that in some areas our cross-sectional profiles possibly do not capture the full extent of OFG bodies. Our OFG volume calculations use a regional coastal length neglecting local and regional curvature. This simplification can lead to OFG volume overestimates as coastline curvatures (e.g. bay and lagoon areas) are not taken into account leading to artificially stretching the coastal length in our calculation. The constant porosity of 0.3 that is used in the groundwater models contributes to underestimating the uncertainty in the global OFG volume. We can deduce that from the sensitivity study with varying porosity values (see SI figure 4 and SI table 3) that shows non-linear decline in estimated OFG volume with lower porosity values. On average, the groundwater models with the lowest porosity value (0.1) show a lower estimated OFG by 11%. This means that the estimated global OFG volume would be lower by 0.11 × 10⁶ km³ than estimated by groundwater models with the constant porosity value equal to 0.3. The differences in estimated mean fresh groundwater fraction between the simulations with low porosity (0.1) and the constant porosity value considered in this study (0.3) can vary from 0.4% up to 34.5% (see SI table 3). Therefore, it is important to take this into account when conducting regional scale studies to investigate the potential impact of varying porosity on the OFG conditions in a given region.

Comparing the estimated OFG volume with current regional water demand highlights areas where OFG exploration could provide a sustainable fresh water source over long periods of time (up to millennia). South-east Asia, West Africa and most of Australia show the highest potential for OFG exploration (figures 3 and 4; SI figures 7 and 8). Exploring OFG as potential fresh water source could be especially interesting in densely populated and intensely farmed deltaic areas (SI table 4). However, more detailed regional hydrogeological and economic analyses are necessary to identify the exact potential the OFG brings to regional usage as source of fresh water. OFG volumes found in several regions could provide multiple centuries (even up to two millennia) worth of fresh groundwater under current and under future water demand rates. However, salinization of inland aquifers might occur when extracting these OFG volumes and this process should be further studied before extracting OFG. This threat is examined by previous studies showing that extracting OFG as an additional source of fresh water for coastal communities is deemed to have potential impacts (Knight et al 2018) on environmental and human conditions while its feasibility should be evaluated for each site (Haakon and Fridtjof 2012).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We also evaluated total OFG volume at shallow depths (higher than 300 m below average sea level; SI figure 9). We estimate the total volume of shallow OFG to be approximately 30% of total OFG volume (∼0.31 × 10⁶ km³), showing still substantial shallow OFG volumes in e.g. South and South-east Asia. The shallow depths are of interest because pumping from deeper formations is more expensive and has a higher potential to disturb onshore groundwater heads (Yu and Michael 2019) in regions where onshore and offshore aquifer layers are interconnected. The connection of the onshore and offshore aquifers is shown to lead to larger OFG volumes via preferential pathways in heterogeneous geological systems as compared to homogeneous systems (Michael et al 2016, Knight et al 2018). Our geological scenarios take into account geological heterogeneity and simulate onshore and offshore interconnectivity (see SI figure 2). However, more local and regional geological information (e.g. boreholes, faults, geological models) is necessary to improve regional scale simulation of OFG.

To conclude, our analysis shows that there are potentially large OFG volumes of up to 10% of total fresh terrestrial groundwater deposited in unconsolidated sediment systems in many regions worldwide. Exploring these offshore resources could provide additional fresh water sources for agricultural, industrial or domestic use. In South-east and East Asia, West Africa and several regions in South America. OFG could prove to be a vital water stress mitigating factor, especially in regions facing near future increases in water demand due to population growth and climate change throughout the 21st century. Further research should investigate the effects of offshore pumping on groundwater levels inland as those could be interlinked. The OFG extraction can potentially lead to onshore groundwater level drops even in shallower aquifers (Yu and Michael 2019) which could have negative impact environmental conditions and societal fresh water needs. Moreover, searching for new hydrogeological data sources, implementing these in local to regional scale hydrogeological modelling studies would help to better understand the extent and behaviour of these offshore fresh (and brackish) groundwater reserves.

This study's calculations were performed on the Dutch national supercomputer Cartesius with the support of SURF. This research was funded by the Netherlands Organization for Scientific Research (NWO) under the STW project 'Water Nexus: Resource Analysis and Regional Water Management' (No. 14298).

The data that support the findings of this study are openly available at the following URL/DOI: 10.5281/zenodo.3937204. The final groundwater salinity estimations for each COSCAT region (as 2D profiles) are provided there as well as shapefiles containing offshore fresh (and brackish) groundwater volumes together with inland water demand and groundwater extraction rates.

D Z, M F P B and G H P O E designed the study and wrote the manuscript. D Z performed the model calculations and data analysis. E H S and L P H (R) v B provided coastal water demand data and helped improve the manuscript.