Groundwater as a major source of dissolved organic matter to Arctic coastal waters

Groundwater is projected to become an increasing source of freshwater and nutrients to the Arctic Ocean as permafrost thaws, yet few studies have quantified groundwater inputs to Arctic coastal waters under contemporary conditions. New measurements along the Alaska Beaufort Sea coast show that dissolved organic carbon and nitrogen (DOC and DON) concentrations in supra-permafrost groundwater (SPGW) near the land-sea interface are up to two orders of magnitude higher than in rivers. This dissolved organic matter (DOM) is sourced from readily leachable organic matter in surface soils and deeper centuries-to millennia-old soils that extend into thawing permafrost. SPGW delivers approximately 400–2100 m3 of freshwater, 14–71 kg of DOC, and 1–4 kg of DON to the coastal ocean per km of shoreline per day during late summer. These substantial fluxes are expected to increase as massive stocks of frozen organic matter in permafrost are liberated in a warming Arctic.


Supplementary Figures
Supplementary Figure 1. Boat-towed measurements of 222 Rn concentrations around Kaktovik Lagoon conducted on August 21 st and 22 nd 2017. Higher 222 Rn concentrations were found adjacent to tall (~ 10 feet) eroding bluffs on the eastern side of Barter Island, where suprapermafrost groundwater (SPGW) may be greater due to steeper hydraulic gradients. Likewise, higher 222 Rn concentrations were found adjacent to a wetland along the southwestern corner of the lagoon, which may deliver relatively large amounts of SPGW during the summer.

Supplementary Note 1.
In this section we provide a step-by-step description of the steady-state Rn isotope box model calculations that were used to estimate total groundwater discharge to Kaktovik Lagoon. This model uses the same mass balance models described in a number of other studies [1][2][3][4] . In particular, our model adopts a framework for lake systems in ref (5).
Assuming that Kaktovik Lagoon is well mixed and that lagoon Rn activity is at a steady state (i.e., dC la /dt = 0), then only the following quantities need to be determined for the model: the mean Rn activity concentration of lagoon water C la-Rn , a representative Rn activity concentration of groundwater inflow C gw , the radium ( 226 Ra) activity concentration of the lagoon water C la-Ra , a representative value for the Rn sediment flux F sed , and the gas exchange flux of Rn from the lagoon surface to the atmosphere F atms . We assume that the lagoon water level remained the same during the sampling period (21 & 21 August 2017) and so we can treat the total lagoon volume (V) and exchange with groundwater as constant. We also assume that lagoon Rn activity concentrations do not change over time. This assumption is reasonable because of the shallow and largely enclosed nature of Kaktovik Lagoon, which is expected to cause a relatively long residence time of water over the summer 6 . Given the assumptions, the Rn mass balance equation is thus: (1) Q gw (C gw − C la-Rn ) + F sed A sed -F atms A surf − λ Rn V (C la-Rn ) + λ Ra V (C la-Ra ) = 0 Q gw is the unknown volumetric groundwater inflow (m 3 d −1 ), which can be calculated when all other inputs, outputs and sinks/sources are known: the input of Rn from lagoon sediments F sed A sed , the loss of lagoon Rn to the atmosphere F atms A surf , the loss of lagoon Rn to natural decay λ Rn V (C la-Rn ), and the introduction of Rn to lagoon water through Ra decay λ Ra V (C la-Ra ).
C gw (Bq m −3 ) was estimated from the average of three groundwater samples (222.5 Bq m −3 ). The discrete groundwater samples were analyzed using 250 ml samples following the Wat250 protocol for the RAD7 radon-in-air monitor (Durridge Company, Inc.). Each Rn measurement was corrected for 222 Rn that has decayed during the time elapsed between sample collection and measurement in the field following: (2) C gw-initial = C gw-final / e −λRn t . The measurements were done in-situ using the RAD-AQUA accessory (or water degassing chamber). Water from approximately 2-3 feet depth was pumped continuously into the RAD-AQUA accessory. The gas in the chamber was sent to three separate RAD7 units following the approach of ref (7). The RAD7 monitors each measured Rn every 30 minutes but each unit was started 10 minutes apart. With the boat moving at 1-2 knots, this allowed for traversing Kaktovik Lagoon over two days. The Rn measurements shown in Supplementary Figure 1 are located at the mid-point of a 30-minute long traverse preceding each Rn measurement.
A surf is the area of the lagoon surface and A sed is the area of the lagoon sediment surface, which are assumed to be the same. A sed and A surf (2.47 × 10 7 m 2 ) were estimated from the perimeter of Kaktovik Lagoon (2.88 × 10 4 m) using the Google Earth polygon tool. A small embayment near a stream outlet on the eastern side of Kaktovik Lagoon was not included in this area estimate. V of Kaktovik Lagoon (6.15 × 10 7 m 3 ) was estimated from A surf and the average lagoon depth (2.5 m), which was calculated from a NOAA bathymetric map.
The other terms for the Rn box model were calculated from the following series of equations: (3) F sed = (D s λ Rn ) 0.5 × (C sed -C la-Rn ) F sed (Bq m 2 day -1 ) is the flux of Rn from lagoon benthic sediments. D s (m 2 day −1 ) is the wet bulk sediment diffusion coefficient. Here we assume a D s value of 4.22 × 10 −5 m 2 day −1 for Kaktovik Lagoon, which is derived from the 154 cm 2 yr −1 estimate for benthic sediments at 0-1 cm depth in Toolik Lake, Alaska reported by ref (9). The diffusive conditions in benthic sediments of Toolik Lake can be postulated to be similar to that in Kaktovik Lagoon because their surrounding environments are very similar. C sed (Bq m −3 ) is the equilibrium activity of Rn measured from wet lagoon benthic sediments. C sed was estimated from the average of seven lagoon sediment samples collected in late August 2017 (2932 Bq m −3 ). Each sediment sample was dried and measured for Rn activity in a bulk emissions chamber. Rn concentrations were then converted to the expected Rn activity from wet lagoon sediments using the equation: (4) C sed = Mn Rn-box / V w C sed (Bq m −3 ) is the expected Rn activity concentration from wet lagoon sediments in the bulk emissions chamber given a sediment porosity (ϕ sed ) of 0.25 and particle density of quartz; this porosity was assumed and is typical of silty sand such as those present in Kaktovik Lagoon. Mn Rn-box (Bq) is the measured Rn production from a known mass of dry sediment. V w (m 3 ) is the volume of water or air in the bulk emissions chamber (2800 cm 3 ).
F atms (Bq m 2 day -1 ) is the loss of Rn to the atmosphere from the lagoon water-air interface. F atms is calculated from the following series of equations: (5) F atms = k(600) × (C la-Rn -α C air ) k(600) (m day −1 ) is the gas transfer coefficient and C air (Bq m −3 ) is the Rn concentration in the air, which is assumed to be 0.10. α is the Ostwald's solubility coefficient. k(600) was estimated using the following series of equations 10 : is the absolute viscosity and ρ (kg m −3 ) is the density of lagoon water with a water temperature of the 9.68 °C and salinity of 20.4 PSU. Temperature and salinity were not measured throughout Kaktovik Lagoon during the study period. Rather, these values were calculated from 15 August 2012 data reported in ref (11). Therefore we assume that these values represent ambient lagoon conditions during our sampling period. In step (7), D m is calculated by:

Supplementary Note 2.
Here we quantify uncertainty in total groundwater discharge by propagating standard errors (SE) for the main components of the Rn box model. These SEs reflect variability between Rn measurements as opposed to analytical uncertainty. Analytical uncertainties for Rn measurements are typically 10-15 % using these methods 4   This analysis demonstrates that we can expect an uncertainty in SPGW discharge and DOM fluxes that ranges from 76 % higher and 59 % lower than the average estimates.

Description of Additional Supplementary Files
File Name: Supplementary Data 1 Description: Project metadata used to generate the figures and tables in the main text and supplementary information are provided in the Supplementary Data 1 file. A description of the data columns is provided on the first tab of the Supplementary Data 1 file.