Evolution of groundwater system in the Pearl River Delta and its adjacent shelf since the late Pleistocene

Abstract

Our extensive field studies demonstrate that saline groundwater inland and freshened groundwater offshore coexist in the same aquifer system in the Pearl River delta and its adjacent shelf. This counterintuitive phenomenon challenges the commonly held assumption that onshore groundwater is typically fresh, while offshore groundwater is saline. To address this knowledge gap, we conduct a series of sophisticated paleo-hydrogeological models to explore the formation mechanism and evolution process of the groundwater system in the inland-shelf systems. Our findings indicate that shelf freshened groundwater has formed during the lowstands since late Pleistocene, while onshore saline groundwater is generated by paleo-seawater intrusion during the Holocene transgression. This reveals that terrestrial and offshore groundwater systems have undergone alternating changes on a geological timescale. The groundwater system exhibits hysteresis responding to paleoclimate changes, with a lag of 7 to 8 thousand years, suggesting that paleoclimatic forcings exert a significantly residual influence on the present-day groundwater system.

The coexistence of inland saline groundwater and continental shelf freshened groundwater is revealed by paleo-hydrogeologic model.

INTRODUCTION

Large-river deltaic estuaries (LDEs) and adjacent continental shelves represent important interfaces between continents and oceans, facilitating the exchange of water and materials (1). They also serve as natural recorders of global environmental changes (2, 3). LDEs typically encompass inland areas such as deltaic plains and lowland floodplains, as well as offshore subaqueous delta that can extend far into the adjacent continental shelf (1). The subaerial zones of LDEs cover a total area of 8.5 × 10⁵ km² globally, accounting for only 0.57% of the earth’s land surface area; however, they are inhabited by over 5% of the world’s population (4, 5). Fresh groundwater resources play a vital role in meeting the increased water demand and alleviating water scarcity in LDEs, especially in the context of rapid population growth and urbanization (6). However, they are highly susceptible to a range of natural and anthropogenic influences (7). Previous hydrogeological studies have shown that coastal deltas were widely affected by marine transgression and regression during the Quaternary (8, 9). For example, the multilayered modeling study conducted at Monterey Bay, California, provides evidence that the fresh-salt interface is still responding to Pleistocene sea-level fluctuations and has not yet achieved equilibrium with the present-day sea-level conditions (10). The intrusion of paleo-saltwater can extend from several to hundreds of kilometers inland, and groundwater salinity can range from brackish to brine (8). Offshore freshened groundwater (OFG) has also been shown to be a global phenomenon in the continental shelves adjacent to the river deltas (11–15). However, there are extremely limited studies on such counterintuitive coexistence of saline groundwater and freshened groundwater in an integrative land-ocean perspective (11, 16).

In this study, we conducted comprehensive investigations in the Pearl River Delta and adjacent shelf (PRD-AS) of the northern South China Sea, using onshore and offshore borehole drilling logs, marine geophysical surveys, multilevel groundwater sampling and monitoring systems, porewater records, chemical and isotopic signals, as well as paleo-bathymetry. The hydrogeochemical analysis of water samples revealed the coexistence of terrestrial brackish to saline groundwater (with salinity up to 25 g liter⁻¹) and OFG (with salinity less than 5 g liter⁻¹) in the adjacent continental shelf (Fig. 1). This phenomenon is counterintuitive because it contradicts the common assumption that onshore groundwater is fresh and offshore groundwater is saline, with coastlines serving as the boundary for terrestrial groundwater systems. Inspired by this phenomenon observed in PRD-AS, we explored global archives on coastal groundwater salinity and found that this special phenomenon exists worldwide (Fig. 2A). Illustrative examples include the Waccamaw River Delta (17), Yangtze River Delta (18), Pearl River Delta (16), Mekong River Delta (19), and Niger River Delta (20), along with their adjacent continental shelves. We believe that the current patterns of groundwater flow systems are insufficient to explain this phenomenon in the land-ocean continuum. Instead, we hypothesize that this counterintuitive phenomenon is a key manifestation of the coastal paleo-hydrogeology, which is influenced by multiple paleoclimatic forcings, such as sea-level changes, variations in precipitation, and sedimentation in the LDEs and their adjacent continental shelves (16).

Fig. 1. Regional setting of the Pearl River Delta and adjacent continental shelf in the northern margin of the South China Sea.

(A) Topographic features of the northern South China Sea and the location of cross section A-B. (B) Distribution of groundwater salinity in the basal aquifer of the Pearl River Delta on the basis of previous hydrogeological investigation (24, 52). The locations of groundwater age samples and multilevel monitoring and sampling wells are indicated by purple triangles and cyan circles, respectively. The paleo-coastlines are shown as cyan dotted lines in the Pearl River Delta. (C) OFG is observed in the subaqueous paleo-delta of the Pearl River 180 km offshore of the estuary. The inferred buried paleochannels in the northern shelf of the South China Sea are shown as dotted gray cyan lines (49).

Fig. 2. Coexistence of saline and freshened groundwater in the same aquifer and exploratory boreholes in the PRD-AS.

(A) Global distribution of coexistence of saline groundwater inland and freshened groundwater offshore in the same aquifer. (B and C) Lithologies and water compositions of exploratory boreholes in the PRD-AS. Borehole positions can be found in Fig. 1, and more borehole information is available in figs. S2 to S4. VSMOW, Vienna Standard Mean Ocean Water. Data of porewater salinity of all boreholes are provided in the Hydroshare data repository (see Data and materials availability).

Subsequently, we developed a sophisticated two-dimensional (2D) paleo-hydrogeologic model to elucidate the spatial distribution and evolution of groundwater salinity in the PRD-AS during Holocene and late Pleistocene periods. Our objective is to enhance understanding of how paleoclimatic forcings, sea-level changes, and sedimentary deposition and erosion processes have influenced the spatiotemporal evolution of the groundwater system. Moreover, we aimed to investigate the response of the associated aquifer-aquitard systems to the geological processes of sedimentation during Milankovitch cycles [~100 thousand years (ka)]. Through the integration of isotopic and geochemical data with modeling experiments, our findings will advance understandings of coastal groundwater systems and pave the way for optimal management of groundwater resources in the PRD and other similar coastal areas worldwide in present-day and in the future.

RESULTS AND DISCUSSION

Spatial distribution of groundwater salinity in the PRD-AS

The subaqueous paleo-delta of the Pearl River, with buried paleochannel systems (see sub-bottom seismic profiles of paleochannels in fig. S1), was developed on the northern shelf of the South China Sea, associated with the deltaic floodplain (Fig. 1A) (21–23). In the subaerial delta, we derived the spatial distribution of groundwater salinity using 113 drilled boreholes, farm wells, and three multilevel groundwater monitoring and sampling wells (refer to the “Sampling sites and field works” section) (8, 24). Brackish groundwater (1 to 10 g liter⁻¹) extends up to 75 km inland from the shoreline, whereas saline groundwater, with salinity >10 g liter⁻¹, is primarily located within 50 km of the coastline (Fig. 1B). The distribution of OFG off the Pearl River Estuary is depicted by salinity profiles obtained from 32 offshore boreholes drilled in the continental shelf of the northern South China Sea [figs. S2 and S3 and the Hydroshare data repository (see Data and materials availability)]. Our boreholes reach depths of up to 303 m below the seafloor (mbsf) and extend as far as 180 km offshore (Fig. 1C and fig. S3), so as to embrace the domain and timescale of Quaternary boundary of the Pearl River subaqueous paleo-delta (23). These comprehensive field data further clarify the potential presence of a substantial OFG reservoir in the continental shelf of the northern South China Sea within a water depth of 100 m.

Specifically, the salinity of groundwater generally increases from the inland deltaic plain to the coastline. The salinity of groundwater in monitoring well SD is less than 2 g liter⁻¹, but salinity increases to 25 g liter⁻¹ near the coastline at well MZ. Both monitoring wells HP and MZ exhibit a similar vertical pattern, where groundwater salinity increases with depth (Fig. 2B). In contrast, offshore groundwater in the intermediate offshore zones (i.e., PR1, 30 km offshore), is only 4 to 5 g liter⁻¹ below 12.5 mbsf (Fig. 2C). Its salinity may even show a decreasing trend with offshore distance and sedimentary depth, as indicated by a minimum salinity of 1 g liter⁻¹ in BH5 (Fig. 2C). This vast OFG reservoir (with salinity of ≤10 g liter⁻¹) is also observed at the depths ranging from 39.4 to 106.7 mbsf in an internationally cooperative Quaternary borehole that we recently drilled (MargS-001) about 180 km offshore the Pearl River Estuary (Fig. 2C and fig. S3).

Evolution of groundwater system in the PRD-AS

With the aim of providing a mechanistic explanation for the counterintuitive distribution of groundwater salinity and establishing the PRD-AS hydrogeological model, we reconstructed paleo-bathymetry and coastlines for the northern shelf of the South China Sea since the Last Glacial Maximum (LGM). This reconstruction was based on extensive paleo–sea-level and age-depth data (figs. S5 to S7 and movie S1) (25). The deposition center is primarily situated outside the present-day Pearl River Estuary (Fig. 3A), due to the abundant sediments transported by the river system (22). From 4 to 10 ka before the present (B.P.), the sea level gradually rose and approached or surpassed the present-day level. This caused the direct flooding of the subaerial delta of the Pearl River during the last transgression (4 to 7 ka B.P.) (Fig. 3, B and C, and fig. S7). Afterward, the sea level experienced a rapid decline during the 10 to 20 ka B.P., resulting in the emergence of submerged continental shelves as terrestrial environments (Fig. 3, D to F) (26). During the LGM, the relative sea level was over 123 m lower than the present-day level (9, 27, 28), and the continental shelf of the northern South China Sea was almost closed to marine influence (Fig. 3F) (29). The exposed continental shelves were covered by freshwater lake and river systems and were subject to infiltration of atmospheric rainfall (30). The groundwater system extended toward the maximum regressive coastlines, with a steeper hydraulic gradient (Fig. 3F, fig. S6, and movie S1).

Fig. 3. Reconstructed paleo-bathymetry and deposited thickness and hydrological model results.

(A to F) Reconstructed paleo-bathymetry, coastline locations, and thickness of deposits at different time slices. See the Supplementary Materials (figs. S6 and S7 and movie S1) for details. (G) Computed present-day mean groundwater ages after the transient simulation. (H) Computed salinity conditions at present-day (the detailed evolution of groundwater salinity can be found in movie S2). (I and J) Simulated present-day hydrogen and oxygen isotopic compositions (δ²H and δ¹⁸O) of groundwater for the cross section A-B. (K) Simulated distribution of groundwater salinity after 100-ka transient simulation, assuming that the climatic conditions (i.e., mean sea level, precipitation) do not changed. Dashed line denotes the present-day sea level. Further information on the model setup can be found in figs. S8 and S9 and table S1.

We further conducted simulations to investigate the evolutions of groundwater salinity, age, δ²H, and δ¹⁸O of the PRD-AS since 50 ka (Fig. 3, G to J) and validated the model with present-day observations (fig. S9). The simulated ages closely corresponded with those obtained from observed ¹⁴C dating (fig. S9B). For example, we found that the shallow low-salinity groundwater in the subaerial delta has ages ranging from 0.21 ± 0.02 to 2.2 ± 0.2 ka B.P., while the deep saline terrestrial groundwater exhibits ages up to 8.8 ± 1.3 to 6.0 ± 0.03 ka B.P. (Figs. 3G and 4A). These findings suggest the terrestrial groundwater in the upper layer has been flushed by rainfall infiltration, whereas the high-salinity groundwater in the PRD region intruded into the basal aquifer during the Holocene transgression. The modeled age framework enables us to extrapolate the ages of the OFGs in the continental shelves, as it is not feasible to obtain sufficient porewater (usually 10 to 30 ml) for groundwater age dating (i.e., ⁴He, ¹⁴C, ³⁶Cl, and ⁸¹Kr) (12). Notably, the modeled OFG ages range from 15 to 30 ka B.P. (Fig. 3G), indicating the OFG formed at the low-stand periods during the last Pleistocene (movie S2).

Fig. 4. Stable isotopic compositions and groundwater ages of water samples collected in the Pearl River Delta.

(A) Present-day oxygen and hydrogen isotopic compositions and groundwater ages in the Pearl River Delta. The shallow groundwater has a younger age and lower salinity, with isotopic compositions approaching meteoric precipitation, while the ages of deeper saline groundwater are close to the early-Holocene period. MSL, mean sea level. (B and C) Modeled changes in the volumes of freshened and brackish groundwater with sea-level fluctuation.

The modeled δ²H and δ¹⁸O show a similar pattern to observed salinity distributions (Fig. 3, H to J), revealing their analogous sourcing and mixing mechanisms. Our model and field observations both illustrated enrichments of δ²H and δ¹⁸O in the terrestrial groundwater system, while the OFG areas show overall depletion. This suggests distinct groundwater sources for the terrestrial and offshore units of the PRD-AS (Figs. 2, B and C, and 3, I and J; and fig. S4B). The δ¹⁸O and δ²H values of the terrestrial shallow groundwater in the PRD-AS closely resemble the isotope content of modern rainfall [−6.29 per mil (‰) and −42.5‰ for δ¹⁸O and δ²H, respectively] (Figs. 2B, 3, I and J, and 4A) (31), indicating that modern rainfall serves as the primary source of groundwater recharge in these areas. The δ¹⁸O values of the terrestrial deeper brackish and saline groundwater approach the isotopic content of the Holocene seawater (32), ranging from −0.8 to 2.0‰ (Figs. 2B, 3J, and 4A). By combining stable isotopic compositions with the terrestrial groundwater age data (Fig. 4A), we found that the enrichment trend in δ²H and δ¹⁸O in the terrestrial deeper groundwater can be attributed to the influence of seawater infiltration during the Holocene transgression (~4 to 8 ka B.P.). The isotopic compositions of the OFG in the continental shelf exhibit high variability, ranging from −0.2 to −5.1‰ for δ¹⁸O and −0.8 to −29.5‰ for δ²H (Figs. 2C and Fig. 3, I and J, and fig. S4). This variability is caused by fluctuations in isotopic compositions of the Pleistocene meteoric water (33).

Last, we also simulated groundwater salinity in the PRD-AS assuming that the climatic conditions (i.e., mean sea level and precipitation) remain consistent with present-day conditions (Fig. 3K). Over a time period of 1 × 10³ ka, the brackish to saline groundwater in the terrestrial areas eventually becomes completely fresh, while the OFG reservoir in the adjacent continental shelf of the northern South China Sea gradually vanishes. This process occurs at a relatively slow rate due to the low permeability of the upper clay or silt layers, which typically have permeability values ranging from 1 × 10⁻⁷ m s⁻¹ to 1 × 10⁻¹¹ m s⁻¹ (8, 9). In conclusion, our comprehensive studies demonstrate that the counterintuitive distribution of groundwater salinity in the present day is the result of interplay among hydro-stratigraphy, sea-level change, sedimentation and erosion processes, and paleo-precipitation variation over geological timescales.

Response of groundwater system to paleoclimate change

The groundwater system can archive clues to paleoclimate changes (34). The volumes of freshened and saline groundwater, defined with thresholds of 1 and 10 g liter⁻¹, respectively, are used as proxies to illustrate how the groundwater system responds to paleoclimate changes. Figure 4 (B and C) compares simulated volumes of freshened and saline groundwater with sea-level changes over the past 50 ka. The volume of freshened groundwater gradually increases (Fig. 4B) as the exposed continental shelf area expands. However, the response of the groundwater system lags behind the sea-level changes by 7 to 8 ka (Fig. 4, B and C). Both hydrological modeling results and field data indicate that the upper clay or silt layers exhibit a high resistance to rainfall and overlying seawater intrusion. The shallow aquitard (i.e., clay or silt layers) also retards the response of the groundwater system in deeper aquifers to the climate change. The paleo-seawater trapped within the onshore fine-grained marine deposits remains in place and gradually spreads to the neighboring aquifers over a period of millennia following marine transgression (24). Our model also provides mechanistic explanations for similar phenomena in other mega low-lying delta plains, such as Red River Delta and Yangtze River Delta (9, 35).

Continental shelves are the submerged fringes of continents and harbor important aquifers beneath the seafloor. However, on a geological timescale, the realm of the terrestrial hydrological cycle has been expanding and contracting as coastlines migrate with the falling and rising of sea levels (11, 36). It has become increasingly evident that the coastline is not a rigid boundary for coastal groundwater systems and that seawater can intrude inland, while fresh groundwater derived from land may be discharged through the seafloor into the ocean or be embedded in sediment and rocks below the present-day seafloor (12, 15, 37). Because continental shelf aquifers underlie areas that are in a continuous state of transition in response to global climate and sea level, coastal groundwater can provide valuable insights into natural variation in the hydrological cycle over thousands of years or even longer (11, 16, 23). We conceptualized the general evolutionary process of the groundwater system in the river deltaic plain and adjacent continental shelf (Fig. 5, A and B). The transition between terrestrial and offshore groundwater reservoirs resembles a “seesaw” mode, influenced by sea-level changes, precipitation, and other geological processes, where the coastline acts as a “fulcrum” in this context. The salinity changes in the onshore and offshore regions of the aquifer system exhibit periodicity with a cycle of approximately 100 ka, primarily driven by sea-level fluctuations, as evidenced by our paleo-hydrogeological models and extensive observations (Fig. 5C). The land-ocean transition mode is also explainable to other observations and puzzles in the LDEs and their adjacent continental shelves (9, 19, 35, 38).

Fig. 5. Schematic illustration of natural variability of hydrological cycle in the large-river deltaic plain and adjacent shelf over geological timescales under the influence of Milankovitch cycles.

(A) Present-day salinity distribution of groundwater in the large-river deltaic plain and adjacent shelf with saline groundwater inland and freshened groundwater offshore in the same aquifer system. (B) Inversed groundwater salinity distribution during the next climate cycle, with freshened groundwater inland but saline groundwater offshore in the same aquifer system. (C) The composited relative sea-level (RSL) curve over the last 450 ka B.P. (28) with the lowstands from hydraulic control model (53). The divided seesaw stages according to the simulation results with coastline acting as a fulcrum in this context for the onshore and offshore groundwater reservoirs.

MATERIALS AND METHODS

The Pearl River Delta is located in the transition zone between the upland landscape of the catchment basin and the deposition center of the northern continental shelf of the South China Sea (Fig. 1A) (26). The receiving basin of the Pearl River covers an area of 9750 km² (excluding the area of Hong Kong), which includes deltaic plains of 5650 km², the estuary (1740 km²), and rocky islands (2360 km²) (39). The thickness of the late Quaternary sediments varies from 10 m to over 40 m (39, 40). Groundwater and river water samples were collected in the subaerial delta of the Pearl River. The northern continental shelf of the South China Sea is orientated approximately in a northeast-to-southwest direction, as it extends from a ridge system that marks the southern end of Taiwan Strait at about 23°N and 119°E to the northeastern coasts of the Leizhou Peninsula and the Hainan Island at about 20°N and 111°E (Fig. 1A). It stretches from the coast of southern China to the shelf break at the 120-m isobath, with an average width of about 200 km (41). Shelf scientific drilling boreholes, with lengths between 30 and 303 m, were mainly collected from the subaqueous paleo-delta of the Pearl River using the Haiyang Dizhi-10 Ocean Drilling Vessel and Specific Drilling Platform (Fig. 1C and figs. S2 and S3).

Sampling sites and field works

Porewater samples were collected from 32 offshore boreholes in the Pearl River Estuary and adjacent continental shelf between May 2002 and August 2023. Recently conducted expeditions include the Integrated Drilling Expedition (May 2021), the Integrated Drilling and Offshore Groundwater Expedition (July 2021), and the Integrated Drilling Expedition of the Subaqueous Paleo-Pearl River Delta-I and II in the northern shelf of the South China Sea (June to August 2023). Most of the offshore boreholes were drilled using the ocean scientific drilling vessel “Haiyang Dizhi-10.” However, boreholes HK4-10 were drilled in earlier projects using a Specific Drilling Platform in the east of the Pearl River Estuary and were analyzed in our previous studies (31, 42, 43). Porewater was collected on board immediately upon recovery using Rhizon samplers (Rhizosphere Research Products, The Netherlands) from the sediments. The Rhizon samplers, consisting of a microporous tube (2.5-mm diameter, 10-cm length) supported by a nylon wire/polyether-ether-ketone (PEEK) wire and connected to PVC tubing with a female Luer lock connector that can be attached to a syringe. A 3.8-mm hole was drilled into the transparent plastic core liner with a drill bit, and the Rhizon filament was inserted into the sediments. Negative pressure was applied to the Rhizon sampler by attaching a 24-ml plastic pulled-back syringe. The sediment cores remained capped throughout this process to minimize ambient gas penetration. The Rhizon sampler, with a sufficiently small tube pore size (0.15 μm), also serves as a filter, removing microbial and colloidal contamination. In addition, bottom seawater was also collected during the expedition using a hydrophore with the assistance of the ship’s crane and filtered through 0.45-μm filters.

Systematic onshore groundwater sampling has been continuously conducted in the PRD from 2006 to 2022. Groundwater samples were collected from domestic wells, piezometers, and farm wells, as well as porewater samples from the sediment cores of drilled boreholes. As part of the comprehensive sampling works, river water in the PRD was also individually sampled in January 2022. Furthermore, three permanent multilevel groundwater sampling and monitoring wells (SD, MZ, and HP) were drilled along the northwest-southeast transect in the PRD (Fig. 1B). Monthly sampling was conducted regularly using a peristaltic pump (Solinst Co.) to sample groundwater between July 2021 and June 2022. All river water and groundwater samples were filtered through 0.45-μm filters in the field before laboratory geochemical analysis.

The 2% HNO₃ solution was added to cation (Na⁺, K⁺, Ca²⁺, and Mg²⁺) samples in the field to prevent precipitation. Split samples were supplemented with 50 μl of 10% HCl and stored in 5-ml brown bottle at −20°C for DOC measurement. The samples for nutrient salts, including NH₄⁺, NO₃⁻, NO₂⁻, PO₄³⁺, and Si measurement, were stored at −20°C until laboratory analysis. Samples for stable isotopes of δ²H and δ¹⁸O were stored in two separate 2-ml bottles without any bubbles. Gas samples were stored in pre-prepared headspace bottles filled with zero gas and 10 μl of saturated HgCl₂. The remaining water samples were stored at 4°C before laboratory analysis.

Geochemical analysis

Water samples for chemical analysis were analyzed in situ for salinity, oxidation reduction potential (ORP), and pH using portable probes [Hanna Instruments (Pty) Ltd.] and subsequently stored at 4°C for further analysis. The values of δ²H and δ¹⁸O were measured with off-axis integrated cavity output spectroscopy and a Triple Isotope Water Analyzer (TIWA-45EP) at the State Key Laboratory of Marine Geology, Tongji University, and the Ratio Mass Spectrometry (Thermo Scientific 253 Plus, Germany) at the Southern University of Science and Technology. The δ²H and δ¹⁸O values were reported relative to the Vienna Standard Mean Ocean Water. The standard deviations of all water samples and standards were less than 0.5 and 0.1‰ for δ²H and δ¹⁸O, respectively. The cations (Na⁺, K⁺, Ca²⁺, and Mg²⁺) and anions (Cl⁻, SO₄²⁻) were analyzed using ionic chromatography (Thermo Scientific Dionex ICS-1100) and inductively coupled plasma optical emission spectrometry, respectively, in the Hydrogeology Lab of Department of Earth Science, The University of Hong Kong. Concentrations of dissolved ammonium, nitrate/nitrite, silicate, and orthophosphate were measured using Flow Injection Analyzer (Lachat Instrument Quickchem 8500) at the Major Equipment Laboratory of the School of Biological Sciences, The University of Hong Kong. The total alkalinity was in situ measured by titrating the groundwater and river water samples (~100 ml) with 0.16 N or 1.6 N H₂SO₄ solution. Porewater was split 5 ml and diluted to 20 ml to test the total alkalinity on board immediately by titrating like the onshore works.

Reconstruction of paleo-shorelines and bathymetry in the PRD-AS

We reconstructed the paleo or prognostic shoreline and bathymetry scenarios for the northern South China Sea (105°E to 116°E, 14°N to 25°N) using a recent digital elevation model (ETOPO_2022), relative sea-level curves, and sediment thickness data. As it is challenging to estimate the thickness of eroded sediments and their impact on shoreline change is relatively minor compared to the effects of eustatic sea-level change (39), the erosion process is not considered in the paleo-bathymetry modeling. Age-depth data from 51 marine sediment cores were compiled from previous investigation reports and published literature (see fig. S5 for boreholes locations). Short cores with ages younger than 20 ka but located at scarce sites were also adopted, and the sediment thickness from their lower parts was calculated on the basis of contemporaneous sedimentation rates from neighboring cores. Different bathymetries related to sedimentation at time intervals of 500 years were reconstructed by interpolation of data from the 51 cores (movie S1).

There is no global relative sea-level curve (27, 44), and the relative sea-level change was also nonuniform in the northern South China Sea. Therefore, six coastal relative sea-level curves (fig. S5) were used to interpret paleo-bathymetry differences and describe spatial sea-level change, taking into account local tectonic processes. Because of the scarcity of relative sea-level curves describing sea-level changes from the LGM to early Holocene along the South China coast, we used the Late Pleistocene comprehensive relative sea-level curve, primarily based on data from the South China coast, to complement the other curves to 20 ka B.P. (45). This method is feasible, because the contribution of eustatic sea-level rise greatly exceeded the local tectonic effects during the complementary time of 20 to 10 ka B.P. (25).

The paleo-bathymetry is reconstructed using present-day digital elevation models (DEM₀) [data from ETOPO_2022 with 30–arc sec resolution database, provided in the Hydroshare data repository (see Data and materials availability)], which include both bathymetry and land relief (25). For the paleo bathymetry reconstruction, DEM expresses the terrain elevations in the PRD-AS area (R) at given time t, while dem_t(r) is a single elevation value at location r in the area R at time t. Their relationship can be written as

$${\text{DEM}}_t=\{{\text{dem}}_t(r)\},r\in R$$ (1)

Paleo-shoreline reconstruction requires two more sets of data. One is the relative sea-level data (RSL) including both eustatic and isostatic components, which was obtained from previous studies (46, 47). Another is data on sediment thickness changes (S) between time t and the present because of sedimentation. For the location r, the elevation value at time t can be expressed by the following equation

$${\text{dem}}_t(r)={\text{dem}}_0(r)-{\text{rsl}}_t(r)+s_t(r),r\in R$$ (2)

where rsl_t(r) describes the sea-level height of the location r in the PRD-AS area at time t, and s_t(r) describes the change of sediment thickness at location r due to sedimentation since time t (accumulated thickness as a negative quantity here). Then, we have

$${\text{DEM}}_t={\text{DEM}}_0-{\text{RSL}}_t+S_t$$ (3)

where RSL_t is the difference in elevations of sea-level between today and time t for the investigated area R, which can be created by spatial interpolation of relative sea level. S_t describes the difference surface caused by sedimentation since time t to present, which can be calculated by interpolation with corresponding data deduced from boreholes and high-resolution seismic profiles.

Paleo-hydrogeologic modeling of the PRD-AS

To simulate the evolution of groundwater salinity, ages, δ²H, and δ¹⁸O distribution throughout the Holocene and late Pleistocene, we used the variable-density groundwater flow modeling code USGS-SEAWAT (48) to set up a 2D model for transect A-B in the PRD-AS region (Fig. 1A) (more information about the paleo-hydrogeological model settings can be found in fig. S8 and table S1). The total simulation period is set to 50 ka B.P. for the following two reasons: (i) the simulation duration needs to be long enough to capture the processes that resulted in the current distribution of freshened and saline groundwater in the PRD-AS and (ii) all measured sedimentary deposits in the subaerial delta of the Pearl River have dating ages younger than 50 ka B.P. (49). The model domain is divided into 840 columns and up to 180 layers with an average thickness of 2 m. The bottom and leftmost columns of the active model domain are assumed to be impermeable and are, therefore, set to be a no-flow boundary. In the offshore domain, the uppermost model cells and the rightmost column are assigned a specified head boundary, which corresponds to the sea-level elevation according to the reconstructed eustatic sea-level curve (28), with a seawater concentration of 35 g liter⁻¹. To account for precipitation recharge variations, a specified flux based on the reconstructed paleo-precipitation is assigned to the top of the onshore domain. The values of the paleo-precipitation are based on the Chinese loess ¹⁰Be data recorded in the Luochuan area, as reported in a prior study (50). Because the PRD area and Luochuan area are both influenced by the East Asian monsoon, the present-day ratio of the annual average rainfall between the two regions is assumed to be similar to that in the past 50 ka to reconstruct the long-term precipitation variation of the PRD during the simulation period. Groundwater age is directly modeled by including a first-order irreversible reaction as an additional factor (51). This was set to zero groundwater age when entering the model domain, and, then, 1 year was added for every year of flow inside the model domain.

The onshore and offshore domains change with the sea-level fluctuation as well as the sedimentation processes. As SEAWAT does not permit changes in spatial domain size over time, “undeposited” cells in each stress period are set as inactive cells to ensure that the flow and solute transport system is free from the influence of undeposited parts. Once a “deposition” cell is added to the model, the parameters and boundary conditions in that cell are activated, converting the cell into a normal active cell (fig. S8 and table S1). The sedimentation rate and thickness during the last 20 ka B.P. can be obtained from the data calculated in the previous section, while the sedimentation thickness during each stress period between 50 and 20 ka B.P. was computed using the averaged sedimentation rate and the ¹⁴C ages of sedimentary deposits. A total of 17 stress periods were implemented to capture both changes in the sedimentation processes and the variations of the sea-level elevation. In each stress period, the active domain is fixed, and the inputs for the temporary model are constant. The description of each stress period and reconstructed paleoclimatic indicators used in the simulated model are described in fig. S8 and table S1.

Supplementary Materials

This PDF file includes:

Supplementary Text S1

Figs. S1 to S9

Table S1

Legends for movies S1 and S2

References

Other Supplementary Material for this manuscript includes the following:

Movies S1 and S2