Infrastructure-based prioritization and cost analysis of offshore freshened groundwater: a Northern Adriatic case study

Offshore Freshened Groundwater (OFG) represents a potential unconventional water resource for water-stressed coastal regions. However, its development is hindered by a lack of systematic, cost-effective methods for regional-scale screening and prioritization, which are necessary before expensive site-specific surveys can be justified. This study addresses this gap by developing and applying an integrated, two-step data-driven framework. First, we present a novel, infrastructure-based prioritization for Euro-Mediterranean OFG sites, ranking them based on regional water-stress and proximity to existing desalination plants and oil and gas platforms. This regional analysis, based solely on openly available data, identifies the Northern Adriatic Sea (offshore Rimini, Italy) as the single “all-criteria-met” location and reveals new, high-potential exploration zones (e.g., southern Spain, Belgium and Malta) where high water stress and infrastructure co-exist. Second, we demonstrate a detailed, quantitative assessment workflow within the Northern Adriatic. This workflow yields two volumetric estimates: a conservative mean volume of 766 km 3 based on confirmed freshwater intervals alone, and a broader mean estimate of 2,413 km 3 that incorporates the overlying interval of undefined salinity, probabilistically constrained using empirical data from 8 wells within the study area. A key observation is the strong geological control, with the thickest OFG accumulations correlating directly with the Po River paleochannel. A preliminary energy-focused cost analysis indicates that the low salinity of OFG markedly reduces reverse-osmosis membrane energy, to roughly 23–56% of that required for seawater across OFG salinities of 1–10 g/L at 75% recovery. Applied in a full cost model, this translates to a total desalination cost 10.6–18.4% lower than seawater over the same range, suggesting OFG may be a cost-competitive unconventional water resource. This integrated framework provides a replicable methodology for stakeholders, bridging the gap from regional prioritization to site-specific feasibility for OFG development.

Coastal aquifers have been described as the nexus of the world’s oceanic and hydrologic ecosystems. They also provide water for over one billion people (Small and Nicholls, 2003). Freshwater resources in coastal areas have been exploited beyond sustainable limits (Werner et al., 2013; Ferguson and Gleeson, 2012), and these systems are under increasing pressure due to rapid population growth, urbanization and climate change (Bates, 2008; Hofste et al., 2019). Offshore freshened groundwater (OFG) is water stored in pores of sediments and fractures of rocks in the sub-seafloor and with a total dissolved solid (TDS) concentration below that of seawater (Micallef et al., 2021). In recent years, there has been mounting attention toward OFG systems due to its global occurrence (Post et al., 2013) and water resource potential as suggested by Bakken et al. (2012), who first proposed their use as a more cost-effective source of raw water for desalination compared to seawater. A report by UN-Water (2020) classified OFG as a potential unconventional water source for coastal regions grappling with heightened water scarcity. Recent studies have identified and modeled large OFG reserves in various locations, including the New Jersey Shelf, USA (Gustafson et al., 2019; Thomas et al., 2019), Canterbury Bight, New Zealand (Micallef et al., 2020; Morgan and Mountjoy, 2022), Pearl River Estuary (Sheng et al., 2023), East China Sea (Thomas et al., 2023) and offshore Malta (Haroon et al., 2021; De Biase et al., 2023). The utility of freshwater stored in aquifers that extend offshore beneath the seafloor has been known to coastal communities for centuries, and documented as far back as 1,000 BC (Person and Micallef, 2022). Moosdorf and Oehler (2017) provide a comprehensive historical review of the various ways in which fresh submarine groundwater discharge has been valued as a water resource for drinking, hygiene, agriculture, fishing, tourism, culture, or ship navigation.

A multidisciplinary review by Micallef et al. (2021) provides a global overview of offshore freshened groundwater (OFG) occurrences. The study, which analyzed over 300 records, shows that OFG is widely distributed across continents worldwide. OFG can be emplaced by meteoric recharge, either at present or during sea-level lowstands (Meisler et al., 1984; Michael et al., 2016), sub-glacial and pro-glacial injection (Person et al., 2017), diagenesis (Kastner et al., 1991) or gas hydrate dissociation (Hesse and Harrison, 1981). The large majority of OFG records are meteoric in origin and located in siliciclastic, passive, non-glaciated margins, within 55 km of the coast in water depths of less than 100 m. Micallef et al. (2021) provided a simple quantitative estimate of OFG resource potential by estimating volumes per km of shoreline, accounting for desalination losses, and projecting the number of years of freshwater supply for the nearest major cities based on current usage. This analysis highlighted Cape Town, Melbourne, and Perth as areas experiencing significant water stress and having the highest OFG resource potential. To move forward from the preliminary resource assessment presented in Micallef et al. (2021), a comprehensive evaluation of OFG viability must consider multiple factors, particularly the technical and economic feasibility of extraction, transport, and treatment compared to alternative water sources such as seawater desalination.

Assessing the technical and economic viability of OFG extraction necessitates a comparative evaluation against alternative sources, with primary focus placed on several critical criteria. Water quality and treatment requirements are key, as the lower salinity of brackish OFG offers a significant cost advantage for desalination compared to seawater (Arroyo and Shirazi, 2012; WateReuse Association, 2012), promoting synergistic development with existing coastal desalination infrastructure (Stein et al., 2016). However, potential contaminants like radium, boron, and arsenic must be managed through appropriate pre-treatment (Stuyfzand and Raat, 2010; Schaefer et al., 2016). Sustainability and resource longevity are paramount, requiring detailed, site-specific modeling to guide extraction rates, prevent seawater breakthrough, and ensure continuous supply (Person et al., 2017). The infrastructure and logistics component, though associated with high capital costs (e.g., offshore well drilling), can benefit from leveraging existing oil and gas industry technologies and infrastructure, such as platforms and pipelines (Bakken et al., 2012; Person et al., 2017). Bakken et al. (2012) estimated OFG production costs at 1.07–1.39 US/m 3, finding this economically competitive with seawater desalination costs of approximately 1.46 US/m 3 for reverse osmosis plants, though they emphasized that costs are highly site-specific and depend critically on raw water salinity and distance from shore. The overall economic feasibility therefore requires site-specific cost assessments to determine whether OFG can be comparable to, or lower than seawater desalination alternatives.

While the criteria for OFG feasibility are established, a systematic, cost-effective workflow for preliminary regional prioritization and risk assessment that integrates readily available datasets remains a critical gap in the literature. To date, only one study has systematically addressed the feasibility of OFG utilization. Bakken et al. (2012) presented a pioneering technical concept for offshore withdrawal, water transport, and treatment, demonstrating that no major technical, economic, or environmental barriers exist to disqualify OFG as a potential drinking water source in coastal regions. Their analysis proposed adapting proven technology from the oil and gas sector (jack-up rigs, subsea installations, and pipeline systems) and the water industry (reverse osmosis treatment) for OFG extraction and processing. The study identified key cost drivers, including raw water salinity and distance from shore, and suggested that OFG could be economically competitive with seawater desalination, particularly when aquifer salinity is low. Furthermore, Bakken et al. (2012) outlined broad criteria for identifying suitable regions, emphasizing areas with limited freshwater access, high desalination dependency, and favorable hydrogeological potential, such as the Middle East—North Africa (MENA) region, southern Europe, and parts of South and Southeast Asia. However, the authors explicitly stated that their feasibility assessment was generic and did not incorporate site-specific data, and they identified the need for detailed regional assessments to refine cost estimates and evaluate local conditions. It is important to distinguish between two related but separate problems: the prediction of OFG occurrence in unsampled regions, which requires detailed geological and geophysical characterization, and the assessment of development feasibility in regions where OFG has already been confirmed, which requires integration of infrastructure, logistical, and economic criteria. This study addresses the latter, building on existing OFG occurrence databases (Giustiniani et al., 2025), and predicted resource volumes (Zamrsky et al., 2022) to develop a practical, infrastructure-based screening framework that can identify where confirmed OFG resources are most amenable to near-term development.

The main objective of this study is therefore twofold: (1) to identify and rank European coastal regions with the highest potential for OFG utilization based on proximity to existing infrastructure (desalination plants, oil and gas platforms) and regional water stress; and (2) to develop and demonstrate a cost-effective, data-driven workflow for preliminary OFG screening and feasibility assessment using the Northern Adriatic Sea as a case study. The Northern Adriatic Sea offers an ideal, high-priority region for this demonstration due to the observed presence of freshwater in offshore boreholes, significant coastal water stress, geologically favorable thick fluvial deposits, and a high density of existing offshore and onshore infrastructure. Finally, we build on the work of Bakken et al. (2012), by conducting a partial and site-specific cost analysis, using energy consumption modelling to compare OFG treatment against both typical and state-of-the-art seawater reverse osmosis benchmarks. This workflow integrates publicly available datasets to evaluate OFG feasibility, providing a replicable framework applicable to other data-rich coastal regions worldwide.

This section outlines the methodology for integrating multiple datasets to assess offshore freshened groundwater feasibility. The approach considers key factors including distance from the coastline, which influences transportation and/or pipeline construction costs; existing infrastructure such as desalination plants for brackish water treatment and oil/gas platforms that could be repurposed for water exploration; and regional water stress. This method of assessment was then applied preliminarily to the Northern Adriatic Basin, where OFG has been identified in several wells, providing an overview of its potential applicability.

Identification and ranking of OFG feasibility factors

To identify suitable sites for Offshore Freshened Groundwater (OFG) extraction in the wider European region, we developed a comprehensive database utilizing QGIS 3.22.4 (Dawson et al., 2022) and ArcGIS Enterprise 11.5 (ArcGIS Enterprise, 2025). The analysis presented in this study relies on several existing datasets that have been integrated to support the feasibility analysis. These are summarized in Table 1 including brief descriptions and references.

The compiled database incorporated several key layers: OFG observation: Boreholes in which freshwater was identified below the seafloor were added from the OFF-SOURCE database (Giustiniani et al., 2025).

Water stress & predicted OFG volumes: Regional water stress data (Hofste et al., 2019) and predicted OFG volume in coastal unconsolidated aquifers [clipped from Zamrsky et al. (2022)] were integrated. These datasets provided an overview of specific regions facing acute water stress within the study area that were also adjacent to offshore regions where large volumes of OFG have been predicted.

Infrastructure proximity: The distances between OFG locations and the nearest point along the present-day shoreline were calculated for areas associated with high water stress (Sayre et al., 2019). Seafloor depths of the OFG observation boreholes were extracted from bathymetry [GEBCO Bathymetric Compilation Group (2023)]. The Oil and Gas Infrastructure Mapping (OGIM) dataset was added and clipped to the study area to determine the proximity of existing offshore infrastructure, such as oil and gas platforms (O'Brien et al., 2024), which could be repurposed for OFG utilization. All platforms were included regardless of their current status. The occurrence of existing desalination plants (EMODnet, 2024) was also included as a critical infrastructure criterion. The desalination plant type and capacity were also recorded in the database.

Based on these datasets, OFG sites were subsequently classified into three feasibility tiers. This classification was determined by assessing whether key cost-feasibility criteria related to infrastructure were met, specifically: Seafloor depths less than 125 m, distances of less than 15 km to the shoreline, the existence of nearby desalination plants, and the presence of offshore platforms. The seafloor depth of 125 m was used as a classification as offshore drilling costs generally increase with water depth; shallow-water operations (here defined as <125 m) are considerably less expensive and technically less demanding than deepwater developments [U.S. Energy Information Administration (EIA), 2016]. A 15 km offshore distance cutoff was selected as it corresponds to the upper range of typical infield pipeline lengths. Such pipelines are routinely installed to connect wells, manifolds, platforms, or subsea facilities to host platforms or nearshore processing hubs. These installations are generally considered technically straightforward and are routinely installed in shallow continental shelf environments (Kaiser, 2020), where OFG is most commonly found. Tier 1: High screening priority was assigned to sites where OFG was observed in wells, all four infrastructure criteria (shallow water depths <125 m, shoreline proximity <15 km, offshore platform, and desalination plant) were present, and the adjacent coastal area exhibited high water stress.

Tier 2: Moderate screening priorit y was assigned to regions where OFG was identified within the shoreline proximity <15 km and shallow water depths <125 m and the adjacent coastal area had high water stress, but only one of the two major infrastructure criteria existed (i.e., either an offshore platform or a desalination plant).

Tier 3: Lowest screening priority was assigned wherever OFG has been identified and the adjacent coastal area had high water stress, but no relevant pre-existing infrastructure was present or water depths were deeper than 125 m.

Northern Adriatic case study

In a study area outlined in Figure 1, located in the Northern Adriatic Sea, we conducted a detailed investigation focusing on the capacity and spatial distribution of OFG, and discussed the results in the context of the onshore-offshore aquifer architecture reported in the region. The offshore part of the study area was defined by the available borehole data and covers approximately 20% of the Northern Adriatic Sea, which was exposed during the Last Glacial Maximum and is therefore likely to host OFG (Corradin et al., 2025). Freshwater intervals have also been identified in several offshore boreholes (Livani et al., 2023). The area also includes the paleo–river basin of the Po River (Verheul et al., 2015). River channel systems typically consist of high permeability conduits, which are commonly associated with OFG systems (Sheng et al., 2023; Thomas et al., 2023; Micallef et al., 2020).

Interpolation of the OFG body

For the assessment of the OFG reservoir, the top levels and thicknesses of fresh- and brackish water intervals in both coastal and offshore aquifers were extracted from 37 borehole datasets (Livani et al., 2023), either from direct statements on the logs or through thorough interpretation of SP and resistivity logs. No direct water sampling was recorded on the legacy logs; reported salinity estimates would have been derived indirectly from resistivity and SP logs, which is standard practice for oil exploration wells. Intervals of saltwater within the fresh and brackish zones were excluded. The fresh interval thicknesses were then measured from the top of fresh water to the base of brackish water to obtain the total OFG interval thickness (Supplementary Table S1). The objective of the ViDEPI wells was hydrocarbon exploration, therefore, salinity data are only available from depths of approximately 200–400 m below the seabed. The thickness of the overlying aquifer interval with unknown salinity was also recorded in our analysis. This interval was recorded to account for the possibility that it may also contain freshwater. Previous OFG studies have shown that OFG commonly occurs in shallower subsurface deposits of less than 300 meters below the seafloor (Micallef et al., 2021). In the Canterbury Bight offshore New Zealand, resistivity anomalies attributed to freshwater occurred directly below the seafloor (Micallef et al., 2020). IODP Expedition 313 wells encountered freshwater within the first 5 m–10 m beneath the seafloor offshore New Jersey (Mountain et al., 2010). In a study of OFG occurrence in the Pearl River Estuary and shelf, several wells showed porewater salinity below 1 PSU starting at depths between 20 m and 30 m below the seafloor (Sheng et al., 2023). These well-documented field observations demonstrate that freshwater in OFG systems frequently occurs within the uppermost tens of meters below the seafloor. Given this pattern, shallow uncharacterized intervals in our boreholes have a significant likelihood of containing freshwater and were therefore incorporated into our Monte Carlo simulations to probabilistically account for their potential contribution to the resource estimate.

To constrain the likely proportion of freshwater within the undefined overlying interval, we analysed a subset of 8 wells within the study area for which the freshwater intervals were explicitly identified in the shallow section. For each well, the depth to the top of the freshwater zone (Z fw) and the depth to the top of the brackish water zone (Z bw) were recorded, and the thickness of the known freshwater interval was calculated as their difference. The fresh fraction of the overlying interval was calculated as the ratio of this known freshwater thickness to the thickness of the undefined overlying interval, i.e., the depth from the seafloor to the top of the known freshwater (equal to Z FW):

Where the known freshwater interval exceeded the thickness of the undefined overlying interval, the fresh fraction was capped at 100%. The full well-by-well calculation using Equation 1 is provided in Supplementary Table S2. Across the 8 wells, this analysis yielded fresh fractions ranging from 14 to 100%, with a mean of 72%. Because the undefined overlying interval is the dominant source of uncertainty in the resource estimate, it was treated as a distinct volumetric component from the confirmed freshwater interval, allowing the two contributions to be quantified and reported separately. These basin-specific, empirically derived values were used to define the minimum, maximum, and mode of the triangular distribution assigned to the undefined overlying interval volume in the Monte Carlo simulation.

First, an interval thickness map was interpolated from the known OFG interval thickness data obtained from 37 boreholes, distributed across the study area, as shown in Figure 1. The interpolation was performed in ArcGIS Pro using Simple Kriging with a Normal Score Transformation (Gaussian Kernels approximation) on a 400 m × 400 m grid in the projected coordinate reference system WGS 1984 Web Mercator (Auxiliary Sphere; EPSG:3857). The variogram model combined a spherical component (range: 165.3 km, partial sill: 0.900) and a Gaussian component (range: 248.0 km, partial sill: 0.078), with a nugget of 0.424 and no anisotropy applied (Supplementary Figure S3.1). The searching neighbourhood used a standard configuration with a maximum of 10 and a minimum of 3 neighbouring points across four sectors. The overlying-interval thickness map (seafloor to top of known freshwater) was generated using the same Simple Kriging approach. Bulk volumes for both interval types were then calculated in ArcGIS Pro using the Zonal Statistics tool: (i) the known freshwater interval and (ii) the undefined overlying interval. Nine near-coastal onshore wells (0.5–5 km landward) were included in the interpolation to account for documented onshore–offshore aquifer connectivity in the region (Campo et al., 2026; Corradin et al., 2025), though only the offshore portion of the interpolated surface was used for volumetric calculations. The volumetric ranges from these two interval maps, combined with the empirically derived fresh fraction constraints described above, provided the input parameters for subsequent Monte Carlo simulations of the total freshwater resource.

Monte Carlo simulation of OFG volume

The total volume of offshore freshened groundwater (OFG) was estimated using a volumetric approach defined as: where V is the OFG volume, A is the areal extent of the area of interest, T is the freshwater interval thickness, and ϕ is the porosity.

Because the freshwater column comprises a confirmed interval and an overlying interval of unknown salinity, the bulk volume (A × T), included in Equation 2, was split into two components that were summed before applying porosity:

A was measured in ArcGIS as a fixed value of 9,343 km 2. Because both T and ϕ have uncertainty ranges, the calculated bulk volumes and porosity used in Equation 3 were treated as probabilistic variables in a Monte Carlo (MC) framework (Metropolis and Ulam, 1949). Triangular distributions were selected because they require only minimum, maximum, and mode values as inputs. The inputs were constrained by borehole data and basin-specific measurements for each variable: interpolation uncertainty bounds for the observed freshwater interval thickness, the empirical fresh fraction analysis of 8 wells for the undefined overlying interval, and core and geophysical log measurements for porosity. This makes them appropriate for data-limited volumetric estimation under uncertainty where a plausible range and central tendency can be constrained but distributional shape cannot.

Comparison of interpolated and observed thickness values at the well locations showed that the interpolation overestimated interval thickness by an average of 10%. To account for this bias, we set the maximum volume of the observed fresh interval to the interpolated bulk volume and the minimum to 90% of that value. The mode was set in the middle at 2,090 km 3. The potential volume of freshwater within the undefined overlying interval was the greatest source of uncertainty. The fresh fraction analysis of 8 wells (Section 2.2.1) yielded fresh fractions ranging from 14 to 100% with a mean of 72%. Despite the minimum value of 14%, we adopt a conservative approach, accounting for the possibility that the overburden interval can be completely fresh or completely saline. Therefore, the minimum bulk volume for the overlying interval was 0 km 3, and the maximum was 7,826 km 3 (i.e., equal to the study area multiplied by the overlying interval thickness). The mode was derived from the product of the mean and the bulk volume (0.72 × 7,826 km 3) yielding a value of 5,635 km 3.

Porosity (ϕ) was assigned a triangular distribution with a minimum of 0.30, a maximum of 0.45, and a mode of 0.35, consistent with empirically measured values of 33–45% reported for the PS3 siliciclastic aquifer system of the Adriatic shelf (Campo et al., 2026), and with effective porosity estimates of 30–40% applied to the same basin by Corradin et al. (2025). All the variables were assigned a triangular distribution type; the values used to constrain the MC simulation are summarized in Table 2.

The MC simulation was implemented in Python with 1,000,000 iterations, sampling from input distributions defined by minimum, maximum, and most-likely (mode) values for (a) the volume from the observed fresh interval, (b) the volume from the undefined overlying interval, and (c) porosity. The resulting distribution of V provided an estimate of OFG volume that explicitly incorporated parameter uncertainty.

Cost analysis

Seawater desalination is an obvious choice to serve as a baseline in an evaluation of the cost effectiveness of OFG. It is a readily available and increasingly used technology for supply augmentation in coastal communities that may potentially be served by OFG. Both desalination and potential OFG sources are also similar in that they require large upfront capital costs (CAPEX), as well as operational and maintenance expenditures (OPEX), once up and running. A range of desalination technologies exist and are used in commercial operations. Of these, reverse osmosis (RO) is currently the most popular such technology in new desalination facilities. It is also the most cost effective of the commercially available technologies. For this reason, it was used as the reference cost.

The economic feasibility of OFG would be a function of its salinity. Energy typically represents between 40 and 50% of the cost of seawater RO (Cosin, 2019; Shokri and Fard, 2023). Energy requirements for desalination of brackish water are between 15 and 50% of those of seawater (Shokri and Fard, 2023), and per unit costs of desalination of brackish water are roughly half to one-third those of seawater (Ullah and Rasul, 2018; Cooley et al., 2019).

OFG could be used as either a substitute for or an addition to seawater feedwater in a desalination facility. As shown in Equation 4, the cost-effectiveness decision criteria for choosing OFG over seawater desalination would be circumstances in which: for which C OFG = cost of OFG supplied water and C SW = cost of desalinated seawater, TC = transportation costs, which are a function of distance, E = energy requirements, which are a function of salinity, depth and distance, A = amortized costs of equipment and/or of adapting existing infrastructure (e.g., conversion of oil rig).

The primary factor affecting operating costs of treatment of OFG would be salinity. In the case of the Northern Adriatic, we assumed a range of salinity levels of 1, 5, and 10 g/L, based on values from the OFF-SOURCE OFG Database. This is in contrast to salinity levels of the Adriatic Sea waters, which average in the range of 29 g/L–38 g/L (Artegiani et al., 1997; Mihanović et al., 2021). This lower salinity allows for both reduced energy costs as well as higher recovery rates. In this study we focus on the difference in energy costs between OFG and seawater.

As the objective of this research was to present a template for evaluation of the potential for use of OFG as a municipal water source, rather than evaluate the conversion of a specific facility, we assumed a large-scale desalination facility. For simulations we looked at a facility with a product capacity of 20 million cubic meters per year (mcm/y). This is significantly larger than the size of desalination facilities currently in use in the region, but comparable in size to large-scale facilities currently active along the Mediterranean. We considered three scenarios for treatment of OFG: Separate—Direct desalination of the OFG at an existing desalination facility separate from existing seawater desalination.

Displacement—Blending of the OFG with seawater, keeping overall output of product water produced by the desalination plant constant at 20 mcm/y.

Addition—Blending of the OFG with seawater, increasing overall output of product water produced by the desalination plant by 10% (2 mcm).

For simulation purposes, for the “Separate” scenario, we assumed a recovery rate of 75%. Seawater RO systems are typically limited to 35–50% recovery due to high osmotic pressure and scaling constraints (Shokri and Fard, 2023). Brackish water RO systems, by contrast, commonly operate at 70–85% recovery, with 75% representing a standard operational target for large-scale plants (Greenlee et al., 2009; Shokri and Fard, 2023). Energy consumption was calculated using LGchem Q + simulation software (LG Water Solutions, 2023). For cost analysis, we used a cost of €0.14 per kWh, representative of current costs for industrial customers in Italy. The simulation software assumes state-of-the-art energy efficiency (2.45 kWh/m 3) for a reverse osmosis plant, using a single-pass system with no need for boron removal (typically done for water or treated wastewater used for irrigation). This is significantly lower than the 3.5 (kWh/m 3) cited in the literature as typical energy consumption for seawater RO plants, as it relates only to specific energy needed to pump feedwater through membranes.

A total of 7 potential OFG sites were classified into three tiers of infrastructure feasibility and are indicated by numbered locations on the map in Figure 2. Tier 1 includes only one location, Adriatic Sea, offshore Rimini Italy. and Tier 3 locations were: Tyrrhenian Sea (Palermo, Sicily), Baltic Sea (Denmark), Bay of Biscay (Bayonne, France), Mediterranean Sea (Béziers, France), Mediterranean Sea (Kerkennah Islands, Tunisia), Gulf of Corinth (Greece). Furthermore, we highlighted areas that have very high water stress and existing desalination and offshore platform infrastructure, which are associated with predictions of substantial OFG reserves (Zamrsky et al., 2022). These areas are located offshore Belgium, Spain and Malta (in yellow on Figure 2). While OFG has not been identified in these regions to our knowledge, they are highlighted in this study since they already possess the infrastructure that can warrant exploration for this potential resource offshore.

The analysis of our database identified the North-Adriatic Sea, off the coast of Rimini (green circle in Figure 2), as an OFG location near a water-stressed area that meets all infrastructure feasibility criteria. Several boreholes in this area indicate low salinity levels (<1 g/L) at shallow seafloor depths (<30 m) within less than 15 km of the shoreline. Additionally, multiple offshore platforms and desalination plants are already situated along the coastline. Figure 3 shows a detailed overview of the database content for the Tier 1 region of the Northern Adriatic Basin. The map shows 4 desalination plants within close range of the sites for a total production capacity of 3,156 m 3 /day based on data from EMODnet (2024). The map also highlights the presence of several oil and gas platforms spread across the study region. Fresh and brackish water was identified in all the wells displayed on the map, which are shaded to reflect the seafloor depth. All of the boreholes are within water depths of up to 70 m.

Interpolation of the OFG body

The interpolation of the OFG interval thickness is displayed on the map in Figure 4. It covers an area of 9,343 km 2. Cross-validation confirms a well-calibrated interpolation, with a mean error of −14.2 m, an RMSE of 152.9 m, and a root mean square standardized error of 0.97. The corresponding kriging standard-error surface (Supplementary Figure S3.2) provides a spatially explicit measure of confidence, with prediction uncertainty lowest where borehole coverage is densest and increasing toward the data-sparse margins. The interpolated OFG thickness map represents the combined thickness of the identified freshwater and brackish-water intervals, along with the overlying interval of unknown salinity, thereby depicting the maximum potential interval containing OFG. Thickness values range from 300 m in the northern sector near Venice to about 500 m along the coast near Rimini. The interval gradually becomes thinner offshore to approximately 170 m at the farthest extent of our study area. The spatial distribution reveals an elongated, channelized trend offshore in the southern part of the study area. The trend correlates with the regional bathymetry and the overlay of the paleochannel of the Po River reveals that the thickest accumulation broadly aligns with the likely position of the proto-riverbed plotted on Figure 4 in green (Verheul et al., 2015).

OFG volume estimation

The Monte Carlo method was applied to estimate the volume of OFG and its related uncertainty in the study area. The simulation, comprising 1,000,000 iterations, yielded a probabilistic distribution for the reservoir’s volume illustrated in Figure 5. Considering the confirmed freshwater interval alone, the Monte Carlo simulation yields a mean volume of 766 km 3 with a standard deviation σ of 67 km 3. When the undefined overlying interval is included, incorporating the empirically derived fresh fraction constraints from 8 wells within the study area (Section 2.2.2, Supplementary Table S2) The mean estimated water storage volume was 2,413 km 3, with a corresponding σ of 640 km 3. This large standard deviation indicates significant uncertainty in the volume estimate, resulting from the undefined interval. To quantify the range of likely outcomes, we identified the 10th (P10) and 90th (P90) percentiles of the volume distribution. The low-side estimate (P10) was 1,516 km 3, meaning there is a 90% probability that the true volume is greater than this value. The high-side estimate (P90) was 3,215 km 3, representing a 10% probability of the true volume exceeding this value. Together, these percentiles define the 80% confidence interval of 1,516–3,215 km 3.

Energy consumption costs

Table 3 presents the specific energy needed to pump water through RO membranes for a range of salinity levels (10, 5, and 1 g/L) relative to those for seawater (with a salinity level of 39 g/L). The figures for OFG are given for and two different recovery rates (40%, typical for seawater, and 75%, typical for brackish water). The energy requirements and associated costs for with a 10 g/L salinity level and 75% recovery rate would be 56% of those for seawater given current state-of-the-art technology. The associated energy costs for membrane processes in desalination of OFG with a salinity of 1 g/L would be just 23% those of seawater. In the case of OFG, there would be an increased need for anti-scalants, but this was estimated to amount to €5,000–10,000 annually, depending on the scale of the plant. This would not significantly affect per unit water costs.

A novel framework for assessing OFG feasibility

This study presented the first systematic, infrastructure-based prioritization of Euro-Mediterranean OFG sites using a newly compiled database. The database analysis demonstrated in the Northern Adriatic Basin addresses a critical knowledge gap by providing a cost-effective, regional-scale screening method that can precede expensive, site-specific field surveys. It should be noted that the cost-effectiveness of this screening approach is contingent on the availability of existing borehole datasets and regional geological data. In the Northern Adriatic, this was facilitated by the legacy of decades of hydrocarbon exploration, which produced a relatively dense network of well logs available through the public ViDEPI database. In regions where such datasets are absent or inaccessible, the applicability of this approach may be more limited, and additional investment in data acquisition may be required before a comparable screening analysis can be conducted. It is worth emphasizing the two-step conceptual structure of this framework. The first step is the regional screening analysis, which constitutes the primary methodological contribution of this study: a novel, infrastructure-based prioritization that identifies where confirmed OFG resources coincide with the logistical and economic conditions most favorable for development. The second step, demonstrated here using the Northern Adriatic, applies a quantitative workflow to the highest-ranked site to illustrate how the framework translates from regional screening to site-specific assessment. These two steps address fundamentally different questions: the first asks where OFG development is most feasible given existing infrastructure and water stress, while the second asks how large the resource is and analyzes the associated desalination costs. The prediction of OFG occurrence in regions without confirmed observations is a separate and more complex undertaking, partially synthesized in our analysis through the inclusion of the Zamrsky et al. (2022) predicted resource volumes.

Our methodology enables stakeholders to evaluate two key aspects of OFG utilization feasibility. First, it provides preliminary estimates of the volume of fresh to brackish water hosted in OFG systems. Second, it assesses the infrastructure and logistical requirements for OFG integration into existing coastal desalination operations. Considering the distance to existing offshore platforms is an approach that builds on the previous study by Bakken et al. (2012), which demonstrated that the overall cost is significantly impacted by the proximity of the resource to the point of use, and by the potential to leverage existing infrastructure. Furthermore, unique offshore factors such as water depth and seafloor sediment introduce engineering challenges and expenses not encountered in onshore scenarios (Bailey and Freedman, 2024). These components have all been included in our database. Due to this complex interplay of factors, the cost-effectiveness of OFG extraction cannot be generalized and must be assessed on a case-by-case basis. Our 3-tier classification provides a useful first-pass screening heuristic that can be translated to other regions where the necessary data is openly available.

While a full environmental impact assessment lies beyond the scope of this study, any future utilization of OFG systems would need to account for several environmental considerations that warrant acknowledgment. First, the degree of hydraulic connectivity between offshore and onshore aquifers, already evidenced in this region (Corradin et al., 2025; Campo et al., 2026), is a critical factor governing environmental risk: where connectivity exists, sustained pumping could reduce onshore groundwater availability, promote seawater intrusion into coastal aquifers, and alter submarine groundwater discharge (SGD) fluxes (Yu and Michael, 2019), with potential knock-on effects for benthic ecosystems and coastal nutrient budgets (Amato et al., 2016), and references therein. Second, subsurface fluid extraction carries a risk of seafloor and onshore subsidence due to volume deficit in the aquifer; this is a particularly sensitive concern in the Northern Adriatic, where natural and anthropogenic subsidence is already a well-documented problem affecting cities such as Venice (Gambolati et al., 1984). Third, the desalination of brackish OFG produces brine effluent that must be managed, although its salinity would be considerably lower than that generated by conventional seawater desalination, reducing the disposal burden. Finally, the construction of offshore wells, pipelines, and treatment infrastructure carries risks of habitat degradation, disruption to faunal migration routes and spawning grounds, and noise pollution during drilling operations (Micallef et al., 2021). These environmental dimensions are not unique to the Northern Adriatic setting, but their site-specific magnitude depends heavily on the 3-D aquifer architecture, the degree of land-sea connectivity, and the extraction rates employed, all of which underscore the importance of detailed hydrogeological characterization before any exploitation programme is advanced.

The Northern Adriatic: a Tier 1 case

Our analysis identifies the Northern Adriatic Sea, offshore Rimini, as the single Tier 1 site, within this screening framework. This top-tier ranking is justified by the convergence of all infrastructure-proxy criteria: high regional water stress, confirmed fresh and brackish water in numerous shallow boreholes, and a high density of both offshore platforms and coastal desalination plants. This “all-criteria-met” scenario presents a compelling case for synergistic development. As suggested by Arroyo and Shirazi (2012), the brackish OFG could be blended with seawater to reduce the energy costs of the existing reverse osmosis plants in Rimini. The oil and gas platforms offer potential for shared logistics or retrofitting, as proposed by earlier studies (Bakken et al., 2012; Person and Micallef, 2022). This resource potential is strongly supported by recent geological studies in the region, which have used seismic and well data to map the offshore continuation of onshore Quaternary aquifers (Giustiniani et al., 2022) and confirmed, through 3D modelling, that glacio-eustatic fluctuations created a layered system with significant potential to host widespread OFG (Corradin et al., 2025). There is also historical evidence of OFG utilization at a man-made platform approximately 500 m outside of the Italian territorial waters. Freshwater was encountered at a depth of 280 meters below the seafloor (Angrilli, 2024). The identification of this prime location validates our regional methodology and justifies its selection for the detailed workflow demonstration discussed in the following section.

An additional outcome of our analysis is the identification of zones of potential interest in southern Spain and along the coast of Belgium (Figure 2, yellow). These regions are not currently known for OFG but possess a combination of high water-stress and extensive infrastructure. This finding supports a strategic shift from reacting to incidental discoveries to proactive exploration targeting in areas with the highest pre-existing economic viability. This approach, especially when combined with geological data-driven predictive models [e.g., Zamrsky et al. (2022)], can significantly de-risk future exploration. The Tier 2 and 3 classifications are similarly valuable, clarifying which infrastructure components are lacking and thereby guiding different development and investment strategies.

OFG volume estimate and geologic controls

Groundwater depletion has been identified as a threat to water security in several regions around the globe (Konikow and Kendy, 2005; Wada et al., 2010). The region directly adjacent to our study area is no exception: in the Venetian-Friulian Plain (VFP), terrestrial groundwater resources are threatened by over-extraction and the unsustainable utilization of surface waters, with reported extraction rates of 48.6 m 3 /s against a recharge rate of only 43.7 m 3 /s, a deficit of approximately 4.9 m 3 /s that places long-term freshwater availability at risk (Zini et al., 2023). Against this backdrop, our mean OFG volume estimate of 2,413 km 3 is a striking finding. To place this regional result into a global perspective, several attempts have been made to estimate the total volume of fresh to brackish water stored offshore using different approaches. Some studies estimate freshwater using a Total Dissolved Solids (TDS) cutoff value of 1 g/L, while others include brackish water, using a TDS cutoff of less than 10 g/L. These are briefly summarized in Table 4.

Our mean estimate for the Northern Adriatic (2,413 km 3) represents a globally significant accumulation. This highlights the Adriatic as an OFG site with high resource potential. This direct comparison of resource scale to regional water demand is a critical, established step in assessing OFG’s potential. A similar approach was used in the global OFG review by Micallef et al. (2021), in which the authors performed simple calculations to estimate the potential years of water supply for high-stress cities like Cape Town, Melbourne, and Perth. The mean estimated volume represents over 4,000 times the annual onshore groundwater deficit and provides a powerful, site-specific validation for the Northern Adriatic. It highlights the OFG’s scale and its significant potential as a strategic, unconventional water resource for the entire region.

A key finding of this workflow is the correlation between the seafloor geomorphology, stratigraphy and OFG distribution. Our interpolation reveals a thick (300 m–500 m) accumulation over a vast area (9,343 km 2) that is not uniform. The observed correlation between the thickest part of the plume, regional bathymetry, and the Po River paleochannel (Figure 4) provides a critical insight into its emplacement. This suggests the proto-riverbed most likely acted as a primary conduit for meteoric recharge and deposition during sea-level lowstands, a widely proposed OFG emplacement mechanism (Micallef et al., 2021; Thomas et al., 2023; Sheng et al., 2023; Micallef et al., 2020) and consistent with the regional depositional history (Corradin et al., 2025). This finding is significant as it provides further evidence that paleochannels are high-priority targets for future OFG exploration, not just in the Adriatic but in similar deltaic settings worldwide.

The hydrogeological architecture of the onshore-offshore aquifer system in this region has been characterized in greater detail by Corradin et al. (2025), who conducted three-dimensional geological and stratigraphic modeling of the Venetian-Friulian Plain (VFP) and its Northern Adriatic shelf extension. Their study confirms that the onshore confined aquifer system comprises 11 to 12 distinct layered aquifers, each averaging approximately 15 m in thickness and separated by 10–40 m thick low-permeability confining units, extending to depths of 500 m below sea level. Crucially, the study demonstrates a high likelihood of onshore–offshore hydraulic connectivity, with sandy aquifer units linked to Quaternary glacio-eustatic cyclothems extending basinward into the Northern Adriatic shelf. Each cyclothem contains a high-permeability coastal sandy deposit capable of hosting freshwater. Direct evidence for offshore freshwater is provided by the Assunta 1 hydrocarbon well, located approximately 3 km from the coastline, which recorded freshwater (salinity below 1%) in the upper 340 m of the sedimentary column. This stratigraphic picture is further corroborated by Campo et al. (2026), who conducted a hydrostratigraphic characterisation of the southern Adriatic shelf offshore Emilia-Romagna, mapping the offshore continuation of the PS3 siliciclastic aquifer system and documenting fresh and brackish groundwater volumes within the uppermost 300 m of the shelf sediments. Their study reports mean porosity values of 38.2 and 37.6% for the upper and lower PS3 aquifer groups respectively, derived from core measurements and geophysical logs, and estimates a total freshwater pore volume of 26.5 km 3 within the first 35 km from the Italian coast. This geological framework strongly supports the interpretation that the freshwater intervals identified in the ViDEPI dataset represent the offshore continuation of an established, regionally extensive confined aquifer system, and provides the stratigraphic and hydrogeological context necessary to justify future targeted geophysical surveys and groundwater flow modeling in this area.

Finally, our workflow successfully quantifies the resource’s uncertainty. The large standard deviation (640 km 3) and wide 80% confidence interval of 1,516–3,215 km 3 are not limitations, but rather crucial outputs. The P10–P90 range reported here constitutes a probabilistic expression of sensitivity to all input uncertainties simultaneously, encompassing the full parameter space explored across 1,000,000 Monte Carlo iterations. The Monte Carlo analysis clearly identifies the overlying interval of unknown salinity as the primary source of this uncertainty. This provides a precise, data-driven justification for future, targeted geophysical surveys [such as marine electromagnetics, as suggested by Micallef et al. (2021)] to characterize this specific interval and de-risk the resource.

Cost analysis

A full cost analysis would require specific data about the exact location of the drilling, depth of groundwater, cost of capital for offshore infrastructure, insurance, etc. This study does not attempt to estimate such costs, though studies have estimated the “at sea” costs of extracting OFG as less than 10% of overall costs (Bakken et al., 2012). This study compares only the cost of desalinating OFG vs. seawater. Commercially active RO plants consume roughly 2.5 kWh–4.0 kWh per cubic meter of water produced (Feo-García et al., 2024; Wang et al., 2025). Of this, the amount needed to pump water through membranes typically represents 50–70% of total energy consumption, with the rest used for intake, pretreatment, post-treatment, and brine discharge stages (Elimelech and Phillip, 2011). Total energy consumption typically represents between 40 and 50% of total cost of seawater RO (Cosin, 2019; Shokri and Fard, 2023). Of these energy costs, the largest element is that needed to apply pressure to pass water through membranes. These costs have been dropping significantly in recent years, due primarily to improvements in energy efficiency, and are roughly half of what they were in the 1990s (Zhou and Tol, 2005).

Table 5 presents an overall cost comparison between OFG and seawater desalination, assuming that membrane processes account for 60% of total energy consumption, that energy costs are 40% of overall costs of seawater desalination, and all other costs (materials, human resources, insurance, etc.) are similar for both OFG and seawater. Under these conditions, the total cost of desalination of OFG with a recovery rate of 75% would be between 10.6 and 18.4% less than that of seawater. If the estimate of “at sea” costs of extracting OFG as less than 10% of overall costs given by Bakken et al. (2012) is accurate, it would indicate that OFG could be cost-competitive relative to seawater desalination.

The economic viability of incorporating OFG into an existing desalination facility would depend on both the scale of the facility and the share of OFG in feed water, as well as the salinity of the OFG. Per unit costs for produced desalinated water are highly dependent on the scale of the facility, as fixed capital expenditures are distributed over the quantity of water produced. Amortized capital costs typically represent roughly half of the total cost of seawater RO plants (Ziolkowska, 2015; Shokri and Fard, 2023). Cost assessments for per unit energy used in this study were based on large-scale facilities active in the Mediterranean. While both the capital and operating costs of extraction and pumping of OFG were not assessed in this analysis, initial findings from analysis of energy costs for the desalination process suggest that OFG may be commercially viable as either a supplement to or an alternative to seawater as a source of potable water. Future studies could estimate capital costs for OFG extraction and evaluate amortized capital for various scale desalination facilities to identify the scale at which OFG could be cost-competitive with seawater desalination.

This study addresses the critical need for a systematic, cost-effective methodology to screen and assess Offshore Freshened Groundwater resources. We successfully developed and applied a novel, infrastructure-based prioritization framework for the Euro-Mediterranean region. This 3-tier classification provides the first-pass, regional-scale feasibility screening. It serves as a tool for stakeholders to de-risk and prioritize investment, identifying the Northern Adriatic as a prime, Tier 1 location and highlighting new, high-potential exploration zones (e.g., southern Spain, Belgium, Malta) where high water stress and existing infrastructure converge.

Our detailed workflow in the Northern Adriatic case study provided two key findings: It demonstrated a globally significant OFG resource, with a conservative mean volume of 766 km 3 based on confirmed freshwater intervals alone, rising considerably to a mean of 2,413 km 3 (P10–P90: 1,516–3,215 km 3) when probabilistically accounting for the overlying interval of undefined salinity.

It identified a strong geological correlation, linking the thickest OFG accumulations to the Po River paleochannel, further highlighting paleochannels as high-priority targets in similar deltaic settings worldwide.

Our cost analysis demonstrated that the low salinity of OFG found in the Northern Adriatic markedly reduces the energy required for reverse-osmosis membrane treatment, to roughly 23–56% of that for conventional seawater desalination, which translates to an estimated 10.6–18.4% reduction in total desalination cost. While this is only a partial cost analysis, and does not account for offshore costs such as drilling and transport to shore, it provides a generalizable framework for such assessments. Our preliminary findings suggest that OFG could be a commercially viable alternative or complement to seawater desalination, although a full economic assessment including capital and extraction costs remains necessary to substantiate this. In summary, this study provides an integrated framework, from regional prioritization to site-specific quantification and economic validation, that bridges the gap between global OFG potential and practical, localized feasibility.