South-to-North Water Diversion stabilizing Beijing’s groundwater levels

The study addresses severe groundwater depletion in the North China Plain and Beijing, where overextraction driven by population growth, economic development, agricultural demand, and climate variability has led to water scarcity, subsidence, and ecological degradation. Beijing’s groundwater levels declined markedly during prolonged droughts (e.g., 1999–2007), exacerbating supply risks for a megacity with per capita water resources far below international scarcity thresholds. The research evaluates whether and to what extent the central route of the South-to-North Water Diversion (SNWD) project, together with climate variability and policy-driven reductions in irrigation, have stabilized or recovered groundwater storage in Beijing. The key objectives are to quantify the contributions of diverted water, precipitation variability, and reduced agricultural pumping to groundwater storage changes since 2006 and to project future trends under different precipitation and groundwater-use scenarios.

Prior work has documented global and regional groundwater depletion using in situ data, GRACE satellite observations, and hydrologic models, highlighting hotspots such as Northwest India, California’s Central Valley, the U.S. High Plains, the Middle East, and the North China Plain. In the NCP, reported depletion rates ranged from ~3.5 km³/yr (1983–1993) based on shallow groundwater levels to ~8.3 km³/yr (2003–2010) from GRACE including deep aquifers. Urban centers worldwide (e.g., Jakarta, Chennai, São Paulo) have experienced severe scarcity and infrastructure impacts tied to groundwater declines. Previous modeling studies for Beijing assessed climate and diversion impacts on groundwater using GW flow or coupled SW/GW models but often omitted comprehensive treatment of climate effects, restrictions on groundwater extraction, and key physical mechanisms, potentially biasing attributions. GRACE-based evidence indicates drying in North China and wetting in South China, motivating inter-basin transfers like the SNWD. However, quantitative attribution of Beijing’s groundwater changes to diversion, precipitation variability, and policy measures had not been comprehensively evaluated before this study.

The study integrates observational analyses with high-resolution hydrologic modeling to attribute and project groundwater storage (GWS) changes in Beijing. Data: Monthly groundwater depth measurements (2008–2019; ~110 wells), specific yield from local bulletins, city-scale water-use statistics by sector and source (2006–2018), reservoir storage, and meteorological forcings. GWS anomalies were derived from in situ groundwater levels using specific yield and analyzed with STL decomposition (Seasonal-Trend decomposition using LOESS) to separate seasonal variability and trends. Modeling: A high-resolution Community Water Model (CWatM) coupled with MODFLOW (CWatM-MODFLOW) simulated surface water–groundwater interactions, water demand, and GWS. The groundwater system was represented with a simplified one-layer confined aquifer for tractability. Recharge representation was adjusted by modifying soil-to-GW percolation using an exponential formulation for hydraulic conductivity, addressing preferential flow dominance observed in the North China Plain. Calibration and validation: The model was calibrated for 2006–2010 and validated for 2010–2014 against observed monthly GWS anomalies using a multi-objective evolutionary algorithm (DEAP), optimizing correlation, precision, and trend fidelity (higher weight on trend). Parameter sets included evapotranspiration, soil depth, preferential flow, heterogeneity, and GW recession coefficients. Uncertainty analyses varied aquifer thickness (100–250 m) and accounted for specific yield variability. Attribution scenarios (2006–2018): - Scenario I (SI): No diversion during 2008–2014 and no diversion for domestic/industrial use during 2015–2018, substituting diverted water with additional groundwater pumping to remove diversion benefits. - Scenario II (SII): SI plus freezing agricultural groundwater use at 2008 levels for 2009–2018 to remove the effect of agricultural pumping reductions. - Scenario III (SIII): SII plus replacing 2008–2018 precipitation with 2000–2018 climatology to remove precipitation variability effects. The STL-decomposed trends from observations and simulations were compared to estimate contributions of diversion, reduced irrigation pumping, and precipitation to cumulative GWS changes. Projections (2019–2030): Four combined scenarios were simulated varying groundwater pumping (G1: maintain 2018 levels; G2: 50% reduction) and precipitation (P1: climatology; P2: wetter climate from RCM HIRHAM5 under CORDEX RCP4.5). The minimum diversion assumed was ≥1 km³/yr to 2030, consistent with planning documents. Additional policy context (population control to ≤23 million by 2030) and potential increases in diversion capacity were considered qualitatively.

- Diversion impact: Water diverted to Beijing reduced cumulative groundwater depletion by approximately 3.6 km³ during 2006–2018, accounting for about 40% of total groundwater storage (GWS) recovery. - Climate contribution: Increased precipitation contributed about 2.7 km³ (≈30%) to GWS recovery over the same period. - Policy/irrigation contribution: Policies reducing agricultural groundwater use contributed about 2.8 km³ (≈30%) to GWS recovery. - Model performance: The coupled CWatM-MODFLOW model reproduced interannual and seasonal GWS variability with R² ≈ 0.85 for the validation period (2010–2014); simulated GWS trend was 131 mm/yr vs observed 144 mm/decade (noting units and uncertainties). - Groundwater use changes: Self-contained well withdrawals declined by ~30% from 1.79 km³ (2006) to 1.25 km³ (2016), driven primarily by reduced agricultural use; water-source well withdrawals declined by ~22% from 0.64 km³ (2006) to 0.30 km³ (2016) due to substitution by diverted water. - Allocation of diverted water (2015–2018 total 5.2 km³): ~6% (~0.3 km³) for treated domestic/industrial use, ~19% (~1.0 km³) for environmental uses, and ~14% (~0.7 km³) stored in Miyun Reservoir, supporting conjunctive management and resilience. - Projections: With at least 1 km³/yr diversion maintained and continued demand management, GWS is likely to continue recovering through 2030 across scenarios; wetter climate and reduced pumping (G2+P2) yield the most rapid recovery. - Water scarcity context: Beijing’s per capita water availability is <100 m³, well below the 1700 m³ scarcity threshold, underscoring the importance of the diversion and demand management.

The findings directly address the research question by quantifying how much of Beijing’s recent groundwater recovery is attributable to the South-to-North Water Diversion relative to precipitation variability and reduced agricultural pumping. Diversions explain roughly 40% of the recovery, with climate and irrigation policies each contributing about 30%. This highlights the effectiveness of engineered inter-basin transfers when combined with demand-side measures in water-scarce megacities. Conjunctive use of surface water and groundwater, strategic reservoir storage, and expanded reclaimed water use have improved resilience to droughts and heatwaves. Projections suggest continued recovery under plausible diversion and policy scenarios, though short-term declines may still occur during climate extremes. The results emphasize the necessity of integrated management, balancing urban supply security with environmental flows and socio-economic impacts in the source region, and sustained monitoring and governance to mitigate adverse effects.

This study provides an integrated attribution of groundwater storage recovery in Beijing to three drivers: inter-basin diversion via the central SNWD (≈40%), increased precipitation (≈30%), and reduced agricultural groundwater use (≈30%). Using a calibrated high-resolution CWatM-MODFLOW framework and observational data, the analysis demonstrates that engineered transfers combined with demand management can stabilize and recover groundwater in severely water-stressed urban regions. Projections indicate that maintaining at least 1 km³/yr diversions and continued pumping reductions will likely sustain recovery through 2030, particularly under wetter climate scenarios. Future work should expand long-term monitoring, refine coupled surface–groundwater–human systems modeling (including more detailed aquifer layering and processes), evaluate managed aquifer recharge opportunities, and assess socio-environmental trade-offs in the source basin to guide adaptive, sustainable water management.

- Model simplifications: The groundwater system was represented with a single confined aquifer layer; some processes (e.g., detailed aquifer heterogeneity, saline intrusion) were simplified or not explicitly modeled. - Storage constraint assumption: Scenarios assumed groundwater depletion was not limited by actual storage, potentially affecting extremes in simulated trends. - Parameter and data uncertainties: Differences between simulated and observed GWS reflect uncertainties in aquifer thickness, spatially variable specific yield, recharge parameterizations, and water-use statistics. - Climate and policy representation: Future climate was represented by selected RCM climatologies, and policy scenarios were stylized (e.g., fixed fractions of 2018 pumping), not capturing full policy dynamics or potential stricter future constraints. - Source region impacts not fully quantified: Socio-environmental impacts in the Danjiangkou source area (resettlement, agricultural restrictions, downstream navigation and ecological flows) were noted but not comprehensively modeled. - Seasonal decomposition and attribution: STL-based trend extraction may be sensitive to parameter choices; attribution assumes additive separability of drivers, which may overlook interactions.