Groundwater discharge as a driver of methane emissions from Arctic lakes

Methane (CH₄) is a potent greenhouse gas contributing about one quarter of radiative forcing from long-lived GHGs. Arctic lakes are major natural CH₄ sources with emissions comparable to northern wetlands, and these emissions are projected to increase 2–3 fold by late 21st century. However, the sensitivity of lake CH₄ emissions to climate change remains uncertain due to limited understanding of drivers of lake CH₄ cycling. While lake emissions are often framed as an imbalance between in-lake production and oxidation, high lake water CH₄ can also be caused by terrestrial CH₄ supplied via groundwater. In the Arctic, abundant wetlands and shallow active layers can funnel high loads of carbon and CH₄ to surface waters. Yet, the magnitude, drivers, and variability of groundwater CH₄ inputs to lakes, especially across space and seasons, are poorly constrained. This study asks to what extent groundwater discharge supplies CH₄ to Arctic lakes and drives atmospheric emissions, and which hydrological, landscape, and seasonal factors control these inputs.

Prior work has focused on in-lake CH₄ production and oxidation as key controls of emissions, but external inputs via groundwater can be substantial, particularly in Arctic systems with wetland-dominated catchments and shallow active layers. Two recent single-lake studies in Alaska (Toolik Lake and a tundra lake on the Yukon–Kuskokwim Delta) used ²²²Rn to show groundwater CH₄ can sustain most summer CH₄ evasion, though groundwater CH₄ concentrations there were sometimes lower than those observed here. Terrestrial carbon inputs via groundwater are known to influence lake carbon cycling, but the specific role of groundwater CH₄ on lake emissions has rarely been addressed. Environmental controls on mire CH₄ production and export (temperature, water table, active layer thickness, topography) suggest strong spatial and temporal variability, highlighting the need for multi-lake, seasonal assessments.

Study region: Ten small lakes (1.8–11.6 ha; mean depth ~2.9 m) in the Torneträsk catchment, Arctic Sweden (discontinuous permafrost), spanning a climatic gradient, were sampled during the ice-free season in 2018–2019. Seasonal campaigns targeted summer (June–July) high flow and autumn (September) base flow. Tracer approach: Groundwater inflows were quantified using radon-222 (²²²Rn; T1/2=3.82 d) as a groundwater tracer due to its enrichment in groundwater and conservative behavior. A ²²²Rn mass-balance under near steady-state (over ~2–3 days, matching ²²²Rn residence time) and well-mixed conditions accounted for sources (groundwater inflow, sediment diffusive flux, inlet streams, in situ ²²⁶Ra-supported production) and sinks (atmospheric evasion, outlet loss, radioactive decay). The governing equation was rearranged to solve for the groundwater ²²²Rn flux Fgw = Qgw·CRn,gw. ²²²Rn residence time was estimated from lake depth, gas transfer velocity, and outlet flushing, supporting the steady-state assumption. Lakes were assumed not to lose water via groundwater (so derived Qgw represents minimum inflow). Measurements: Lake surface and depth-profile waters, inlet and outlet streams, and shoreline mire groundwater (20–40 cm depth) were sampled. ²²²Rn in lake/stream water was measured with a RAD7 setup; groundwater ²²²Rn by liquid scintillation (Quantulus 1220). ²²⁶Ra-supported ²²²Rn was determined from large-volume Ra extractions and gamma counting. Dissolved CH₄ in gas-tight vials was analyzed by GC with standards; atmospheric-equilibrium values were assumed where below detection. Physicochemical parameters (temperature, DO, conductivity), stream discharges (electromagnetic meter or salt slug), meteorology (wind, rainfall, air T/P), and sediment properties were collected. Sediment incubations constrained sediment ²²²Rn diffusive flux and provided an independent CRn,gw check. Bathymetry and catchments were derived from echo sounding and a 2-m DEM; catchment properties (slope, wet areas/open mires, soil moisture proxy) were mapped. Flux calculations and uncertainty: Atmospheric fluxes of ²²²Rn and CH₄ used a wind-based gas transfer model (Klaus & Vachon) with sensitivity to alternative parameterizations. A deterministic error propagation for the ²²²Rn mass balance yielded uncertainties in Fgw. Monte Carlo simulations (n=1000 per lake-season) randomly sampled Fgw, CRn,gw, and CRn,gw/CCH4,gw from groundwater observations (n=41) to derive distributions of Qgw and associated groundwater CH₄ inputs; medians with IQRs are reported. Emissions comparison: Diffusive CH₄ emissions were measured; ebullition was not measured directly. To estimate maximum total emissions conservatively, ebullition was inferred assuming diffusion represents 17% (based on Stordalen mire lakes, 9-year dataset), with ebullition contributing the remainder; resulting totals were compared to groundwater CH₄ inputs. Statistical analyses: Differences among waters and seasons were tested with ANOVA and Tukey–Kramer HSD (p<0.05). Spatial drivers of groundwater inflows were explored via partial least squares regression (PLS) using catchment and lake variables plus precipitation; variable importance (VIP) guided selection. Multiple stepwise linear regression identified parsimonious predictors, evaluated by adjusted R² and AIC, checking multicollinearity and transformations. All analyses were conducted in R (v4.1.0).

- Groundwater is highly enriched in CH₄ and ²²²Rn relative to surface waters: groundwater CH₄ median 150 µM (IQR 49–210 µM) vs lakes 0.19 µM (IQR 0.02–0.48) and inlet streams 0.02 µM (IQR 0.02–0.37). No seasonal difference in groundwater CH₄ (ANOVA F=0.50, p=0.48). - Groundwater ²²²Rn median 3500 Bq m⁻³ (IQR 2100–8800) vs lakes 110 (78–160) and inlet streams 520 (260–2100). No seasonal difference (ANOVA F=0.087, p=0.77). - ²²²Rn mass balance indicates groundwater is an important water source for most lakes, with median inflows from 0.18 to 6.4 cm d⁻¹. Inflows higher in summer (1.6–6.4 cm d⁻¹) than autumn (0.18–3.4 cm d⁻¹). Normalized inlet discharges: 0.69 cm d⁻¹ (IQR 0.20–3.0). - Groundwater CH₄ inputs dominate external inputs: 28–120 mg CH₄ m⁻² d⁻¹ (summer) and 2.0–59 (autumn), up to an order of magnitude higher than inlet streams (<0.01–1.3 mg CH₄ m⁻² d⁻¹). - Lake CH₄ emissions: diffusive fluxes 0.70–7.6 mg CH₄ m⁻² d⁻¹ (summer) and <0.01–2.3 (autumn). Estimated ebullition (assuming diffusion=17% of total) <0.1–37 mg CH₄ m⁻² d⁻¹; maximum total emissions 4.1–44 mg CH₄ m⁻² d⁻¹ (summer) and <0.1–13 (autumn). Groundwater CH₄ inputs are of the same order as total emissions, implying groundwater can match/sustain lake emissions regionally. - Summer relationship: total CH₄ emissions scale positively with groundwater inflow (R²=0.61, df=8, F=15, p<0.005; y=(6.5±1.7)x+(0.1±6.2)). No such correlation in autumn. - Spatial drivers: PLS with catchment and lake variables plus precipitation explained 72% of variability in groundwater inflows. Best multiple regression (explained 45% variance): Qgw = 7.3 + 3.5·depth − 5.7·log₁₀(wetzone) − 0.27·log₁₀(slope) (df=20, F=7.2, p<0.002). Inflows increased with lake depth; unexpectedly decreased with greater wet zone/mire cover and with catchment slope. - Seasonal controls: Higher summer inflows consistent with hydrological cycle (snowmelt recharge, higher early-summer flows). Biological controls (temperature-dependent mire CH₄ production) likely contribute to seasonal patterns in CH₄ available for export, though measured groundwater CH₄ concentrations did not differ seasonally, possibly due to residence-time effects. - Magnitude comparison: Groundwater CH₄ inputs are comparable to in-lake production and oxidation rates reported across the Arctic, underscoring groundwater discharge as a major mechanism in Arctic lake CH₄ cycling.

The study demonstrates that groundwater discharge is a pervasive, external CH₄ source to Arctic lakes that can sustain total atmospheric CH₄ emissions during the ice-free season. By quantifying groundwater inflows with a ²²²Rn mass balance across multiple lakes and seasons, the work addresses the uncertainty surrounding terrestrial CH₄ contributions to lake emissions. The strong summer correlation between groundwater inflow and total emissions indicates that hydrological connectivity and recharge dominate emission variability when flows are high, whereas this linkage weakens in autumn under baseflow conditions. Spatial patterns in inflow reflect lake-catchment attributes, notably lake depth (positive effect) and wet zone/mire extent and slope (negative effects), highlighting the interplay between geomorphology, hydrological gradients, and groundwater residence times. Comparing groundwater inputs to other flux components (sediment production, oxidation, diffusion, ebullition) shows similar magnitudes, emphasizing that models of lake CH₄ cycling and regional to global CH₄ budgets should incorporate groundwater pathways. The findings are particularly relevant for the Arctic, where climate-driven changes in permafrost, hydrology, and temperature likely amplify groundwater-mediated CH₄ export to lakes and subsequent atmospheric emissions.

Groundwater discharge is a key, previously underappreciated driver of CH₄ emissions from Arctic lakes. Across 10 lakes in Arctic Sweden, groundwater CH₄ inputs frequently matched the magnitude of total lake emissions, with inflows controlled by lake depth, wetland cover, catchment slope, and seasonal hydrological and biological dynamics. Recognizing and quantifying groundwater pathways improves mechanistic understanding and predictive capacity for Arctic lake CH₄ emissions, especially under climate change scenarios featuring permafrost thaw, warmer temperatures, and increased precipitation that are poised to enhance groundwater recharge and CH₄ production. Future work should (1) directly measure ebullition alongside diffusion and groundwater inputs across broader spatial and temporal scales, including shoulder seasons and under ice; (2) refine groundwater flowpath characterization (e.g., permafrost distribution, hydraulic conductivity, preferential flow) to improve spatial predictions; (3) integrate groundwater CH₄ fluxes into Earth system and regional biogeochemical models; and (4) assess feedbacks between landscape evolution (tundra greening, active layer deepening) and groundwater-mediated CH₄ export.

- Ebullition was not measured directly; total emissions were estimated using an external dataset (Stordalen lakes) assuming diffusion constitutes 17% of total, likely providing conservative, maximum ebullition estimates but introducing site-representation uncertainty. - The ²²²Rn mass-balance assumes near steady-state over ~2–3 days, well-mixed water columns, and no groundwater outflow from lakes; the latter means groundwater inflows are minimum estimates. Conceptual and parameter uncertainties in tracer budgets and gas transfer velocity parameterizations remain. - Sampling covered summer and autumn of two years; winter/under-ice and spring transitions were not directly measured, limiting full annual characterization. - Spatial inference is based on 10 lakes within one Arctic catchment; broader generalization requires larger, more diverse datasets. Complex hydrogeological controls (permafrost distribution, heterogeneity, preferential flowpaths) were not explicitly resolved. - Some CH₄ concentrations were at or below detection and set to atmospheric equilibrium; ebullition and oxidation dynamics were not resolved at high temporal frequency.