1. IntroductionAnthropogenic warming is projected to increase Arctic precipitation, as sea surface warming and sea ice retreat increase local moisture availability and hemispheric warming intensifies remote moisture transport from lower latitudes (Bintanja & Selten,  2014; Singh et  al.,  2017). However, constraining the long-term sensitivity of these mechanisms to increase in regional and global temperatures remains challenging, as

Abstract Increased precipitation in the Arctic is a robust feature across model simulations of the coming century, driven by intensification of meridional moisture transport and enhanced local evaporation in the absence of sea ice. These mechanisms are associated with distinct, seasonal, spatial, and, likely, precipitation isotope (δ2HPrecip) expressions. Historical observations of δ2HPrecip reveal a contrast in seasonality between southwestern and northwestern coastal Greenland: δ2HPrecip in northwestern Greenland varies in phase with local temperature, whereas δ2HPrecip in southwestern Greenland is decoupled from local temperature and exhibits little seasonal variation. We test the hypothesis that reduced δ2HPrecip seasonality in southwestern Greenland relative to northwestern Greenland results from dynamic moisture source variations, by diagnosing monthly average moisture sources to three sink regions (Kangilinnguit, Ilulissat, and Qaanaaq) using the Water Accounting Model-2layers model. All domains demonstrate strong intra-annual moisture source variations. Moisture to the southernmost region is sourced most remotely in summer and most locally in winter, associated with stronger cooling from the source in summer than winter, promoting more negative δ2HPrecip and counteracting local temperature-driven seasonality. In comparison, moisture transport distance to the northernmost region is relatively constant, as local sea ice restricts northward migration of the winter moisture source. We simulate seasonal patterns in δ2HPrecip in a simple Rayleigh model, which confirm the importance of source temperature and starting isotopic compositions in determining δ2HPrecip for these regions. δ2HPrecip sensitivity to moisture source variability suggests these coastal Arctic settings may yield paleoclimate records sensitive to the moisture transport processes predicted to amplify future precipitation.

Plain Language Summary Climate models predict that Arctic precipitation will increase in the future in response to anthropogenic warming, due to increases in local evaporation and moisture transport from low-latitudes. Precipitation isotopes are a powerful tool for understanding current and past hydrological processes, and can be useful tracers of moisture source and transport. In the Arctic, records of precipitation isotopes at high elevation on the Greenland Ice Sheet have traditionally been interpreted in terms of local temperature. Precipitation isotope records are increasingly generated from low-elevation coastal Arctic settings, where local temperature-isotope relationships are not robust. We investigate the role of seasonal moisture source variability on seasonal precipitation isotope patterns in coastal western Greenland, using a model to track precipitation back to its source of evaporation. We find that moisture source variations explain a lack of seasonality in precipitation isotopes in southern Greenland, with moisture source variations acting in opposition to the influence of local temperature. These results indicate that precipitation isotope records from these regions may be sensitive to past moisture source changes, and therefore useful for understanding the relative importance of processes which may impact Arctic precipitation in the future.

CLUETT ET AL.

© 2021. American Geophysical Union. All Rights Reserved.

Seasonal Variations in Moisture Origin Explain Spatial Contrast in Precipitation Isotope Seasonality on Coastal Western GreenlandA. A. Cluett1 , E. K. Thomas1 , S. M. Evans2 , and P. W. Keys3

1Department of Geology, University at Buffalo, State University of New York, Buffalo, NY, USA, 2Department of Geography, University at Buffalo, State University of New York, Buffalo, NY, USA, 3School of Global Environmental Sustainability, Colorado State University, Fort Collins, CO, USA

Key Points:• Seasonal patterns in moisture

transport distance dampen precipitation isotope seasonality in southwestern Greenland

• Sea ice suppresses local winter moisture to northwestern Greenland, amplifying precipitation isotope seasonality

• Moisture transport in a warmer world will likely dampen precipitation isotope seasonality in coastal Arctic regions

Supporting Information:Supporting Information may be found in the online version of this article.

Correspondence to:A. A. Cluett,aacluett@buffalo.edu

Citation:Cluett, A. A., Thomas, E. K., Evans, S. M., & Keys, P. W. (2021). Seasonal variations in moisture origin explain spatial contrast in precipitation isotope seasonality on coastal western Greenland. Journal of Geophysical Research: Atmospheres, 126, e2020JD033543. https://doi.org/10.1029/2020JD033543

Received 16 JUL 2020Accepted 16 MAY 2021

10.1029/2020JD033543RESEARCH ARTICLE

1 of 15

Journal of Geophysical Research: Atmospheres

instrumental records of Arctic precipitation are short, sparse, and characterized by strong interannual var-iability with respect to the expected changes (Kattsov & Walsh,  2000; Linderholm et  al.,  2018; Mernild et al., 2015; Rawlins et al., 2010). Water isotopologues are useful tracers of current and past atmospheric and hydrological processes, integrating the impacts of mass-dependent fractionations between evaporation and eventual precipitation (Bowen et al., 2019). Glacial ice and lake sediments archive direct and proxy records of past precipitation isotopic variability through the late Quaternary, providing an opportunity to study aspects of precipitation changes on multi-decadal and longer timescales (Johnsen et al., 2001; Leng & Marshall, 2004). In this study, we demonstrate the importance of moisture source variations on seasonal coastal Arctic precipitation and its isotopic composition, to guide dynamically informed interpretations of precipitation isotope proxy records from these regions.

Western Greenland provides an excellent case study on the sensitivity of coastal precipitation isotopes to moisture transport processes and source variations, as it is a gateway of remote moisture transport to the Arctic and local moisture availability is modulated by the presence of seasonal sea ice in the Labrador Sea (Dufour et al., 2016). In the Arctic, local temperature is a dominant control on variations in the stable oxy-gen and hydrogen isotopic compositions (δ18O and δ2H) of precipitation, established following observations of spatial isotope-temperature (δ-T) relationships globally and across Greenland (Dansgaard, 1964; Johnsen et al., 1989). Yet, multi-year monitoring of precipitation isotopes from two sites on coastal western Green-land demonstrate contrasting seasonal patterns, revealing that precipitation isotopes do not consistently co-vary with temperature on monthly timescales (Figures 1 and S1) (Bowen, 2008; IAEA/WMO, 2015). In northwestern Greenland, strong seasonal variations in precipitation δ18O and δ2H mirror seasonal varia-tions in temperature, resulting in strong δ-T covariation as expected for a “typical” high-latitude site (Bow-en, 2008). However, in southwestern Greenland, negligible seasonal variations in precipitation δ18O and δ2H yield weak δ-T covariation (Bonne et al., 2014). Muted seasonality in southern Greenland precipita-tion δ18O relative to observed atmospheric vapor δ18O has been partially attributed to the temperature-de-pendence of fractionation between vapor to liquid or solid precipitation (Bonne et al., 2014). However, this

CLUETT ET AL.

10.1029/2020JD033543

2 of 15

Figure 1. (a) Map of sink regions and locations of meteorological stations. Stations are coastal, whereas sink regions include some higher-elevations on the ice sheet. Dotted line indicates backtracking boundary. (b) Monthly mean temperature and interannual standard deviation from Global Historical Climatology Network (GHCN; NOAA, 2020) stations (black) and ERA-interim 2 m-sink region temperature (gray). (c) Monthly precipitation δ2H and interannual standard deviation measured at IAEA Global Network of Isotopes in Precipitation (GNIP; WMO/IAEA, 2015) stations in Kangilinnguit (formerly Grønnedal) and Qaanaaq (formerly Thule) (solid blue) and modeled by the Online Isotopes of Precipitation calculator for all sites (OIPC, dashed; Bowen, 2019). No long-term precipitation isotope monitoring has been undertaken in central western Greenland. Although precipitation δ2H seasonality is poorly constrained for this intermediate site, few measurements of precipitation isotopes (blue points) (Lindborg et al., 2016) suggest intermediate seasonality between northern and southern sites in line with OIPC interpolation. (d) Monthly precipitation amount measured at GHCN stations (boxplots) with median total tracked evaporation marked by black points.

Journal of Geophysical Research: Atmospheres

thermodynamic control can only compensate for a fraction of the expected seasonality, suggesting dynamic controls on moisture source or transport are also important.

At monthly timescales, variations in precipitation δ18O and δ2H in phase with local temperature changes are often observed at high-latitude sites, which are characterized by strong temperature seasonality in com-parison to their lower-latitude moisture sources (Bowen, 2008; Dufour et al., 2016; Hendricks et al., 2000). The expected covariation of precipitation δ18O or δ2H and local temperatures in the Arctic can be simulated using a simple one-dimensional Rayleigh-type distillation model, in which an air parcel cooling from a site of evaporation undergoes preferential rainout of heavy isotopologues (i.e., 2H1H16O, 1H2

18O), driving progressive 18O- and 2H-depletion of the remaining vapor and resulting precipitation (Ciais & Jouzel, 1994; Dansgaard, 1964; Johnsen et al., 1989). Assuming no mixing with other air masses, the isotopic composition of precipitation at a given point along the trajectory of air parcel cooling is predominantly determined by the initial vapor isotopic composition and the fraction of moisture remaining at the time of rainout (Ro-zanski et al., 1993). As temperature determines saturation vapor pressure, the magnitude of cooling of the lofted air parcel between the site of evaporation and the site of precipitation determines the extent of prior rainout and is a main control on the resulting precipitation isotopic composition. Consequently, robust rela-tionships between local temperatures and local precipitation δ18O or δ2H emerge at locations and timescales for which moisture source temperatures are relatively constant in comparison to local temperatures at the site of precipitation (Putman et al., 2017).

Yet, even in settings where these criteria are generally well met, such as on the summit of the Greenland Ice Sheet, paleoclimate and modeling studies indicate that relationships between local temperatures and precipitation δ2H or δ18O vary through time (Jouzel et al., 1997). For example, temperatures inferred from Greenland ice cores based on spatial δ-T relationships yield temperature changes through the last deglaci-ation approximately half the magnitude of estimates derived independently from thermally fractionated gases and borehole temperatures (Buizert et al., 2014; Cuffey et al., 1995; Severinghaus et al., 1998). Modern Lagrangian trajectory analysis and water-tagging experiments in isotope-enabled general circulation mod-els indicate that substantial seasonal and interannual variability in moisture sources to the Greenland Ice Sheet arises from variability in atmospheric circulation and sea surface conditions, and acts as a secondary control on precipitation δ18O and δ2H (Nusbaumer et  al.,  2019; Sodemann et  al.,  2008a,  2008b; Werner et al., 2001). However, despite evidence for an influence of moisture source variations on δ-T relationships, local temperature remains the dominant control on precipitation δ18O and δ2H for the ice sheet setting due to the strength of distillation at low temperatures (Jouzel et al., 1997).

The impact of moisture source variations on precipitation δ18O and δ2H is amplified in coastal Arctic re-gions, which are characterized by warmer moisture transport conditions and enhanced low-level moisture availability relative to the Greenland Ice Sheet summit (Mernild et al., 2015). Observations from coastal Alaska and Greenland demonstrate that event-scale precipitation isotopic variability can decouple from temperature variability in response to dynamic variations in moisture source location and transport path-way (Bonne et al., 2014; Klein et al., 2015; Klein & Welker, 2016; Putman et al., 2017). Indeed, the impor-tance of seasonal moisture source variations on precipitation δ18O and δ2H has been established for low- and mid-latitude sites (Levin et al., 2009; Li & Garzione, 2017); however, the impact of moisture source on precipitation isotope seasonality has not been as fully examined at high-latitudes. A gap in understanding remains regarding the seasonality of moisture sources and their impact on precipitation isotope seasonality on coastal Greenland, where several lacustrine isotope proxy records document changes in precipitation δ2H and δ18O through the Holocene (Balascio et al., 2018; Corcoran et al., 2021; Lasher et al., 2017; Thomas et al., 2016, 2018, 2020), in some cases deviating from records of local and regional temperature (Axford et al., 2021).

Here, we investigate the role of seasonal variability in moisture source locations and temperatures in deter-mining modern and future precipitation isotope seasonality on coastal western Greenland. First, we test the hypothesis that stronger seasonal variations in moisture sources reduce the seasonality of precipitation δ2H in southwestern Greenland relative to northwestern Greenland. Second, we test the hypothesis that future increases in remote moisture transport during the summer and local evaporation in the winter will impact precipitation isotope seasonality in both regions in the future. To test these hypotheses, we compare mod-ern monthly moisture sources to three domains along western Greenland using an Eulerian precipitation

CLUETT ET AL.

10.1029/2020JD033543

3 of 15

Journal of Geophysical Research: Atmospheres

backtracking model, and simulate seasonal patterns in precipitation δ2H in a simple Rayleigh distillation model accounting for changes in moisture source conditions based on the seasonal source region changes determined by the precipitation backtracking model.

2. Methods2.1. Precipitation Isotope Data

We compare multi-year monthly average precipitation δ2H records from northwestern (Qaanaaq; 77.47°N, 69.23°W) and southwestern Greenland (Kangilinnguit; 61.23°N, 48.10°W) from the Global Network of Isotopes in Precipitation (GNIP) (Figure  1) (IAEA/WMO,  2015). No long-term constraints on precipita-tion isotope seasonality currently exist from coastal central western Greenland, although a few available measurements of event-scale precipitation δ2H are in line with moderate seasonality (Lindborg et al., 2016). We additionally compare measured precipitation δ2H values with monthly values simulated for Qaanaaq, Ilulissat (69.22°N, 51.05°W), and Kangilinnguit by the Online Isotopes of Precipitation calculator (OIPC) (Bowen, 2019; Bowen et al., 2005).

2.2. Precipitation Backtracking Model

We diagnose the monthly average moisture sources to three western Greenland domains using the water accounting model configured with two atmospheric layers (WAM-2layers) of van der Ent et al. (2010, 2013, 2014), with the precipitation backtracking analysis developed by Keys et al. (2012). We calculate monthly moisture sources for three 3° latitude by 3° longitude sink regions centered on Kan-gilinnguit, Ilulissat, and Qaanaaq (Figure 1a). We selected these domains to include the sites of modern precipitation isotope measurements from northwestern and southwestern Greenland (IAEA/WMO, 2015), and sites of long-term instrumental precipitation amount records in all domains, against which we validate moisture tracking results (NOAA, 2020) (Figure S2; Text S1). These domains span meridional gradients in temperature, precipitation amount, and proximity to seasonal sea ice (Figures 1b and 1d).

We use the WAM-2layers to track precipitation tagged within designated sink regions backwards in time to its sites of evaporation, based on three-dimensional, model-level data from the ECMWF ERA-Interim Reanalysis (Berrisford et al., 2011). At each time step, the model calculates the water balance of upper and lower atmospheric layers for each grid cell, with evaporation entering and precipitation exiting the column of air and moisture advected horizontally and vertically between layers according to winds. The Eulerian approach of the WAM-2layers backtracking resolves precipitation and evaporation within atmospheric lay-ers through time, differing technically from Lagrangian backtracking techniques in which moisture sources are evaluated based on relative humidity variations, and thus net surface interactions, along isolated trajec-tories (van der Ent et al., 2013).

The WAM-2layers uses specific humidity, zonal and meridional wind speeds at 24 pressure levels (175–1,000 hPa), and surface pressure at 6-h intervals, and 3-h accumulated precipitation and evaporation from the ECMWF ERA-Interim Reanalysis at 1.5° longitude by 1.5° latitude resolution for 1979–2014 (Berrisford et al., 2011). The WAM-2layers integrates between each of the model levels of the three-dimensional rea-nalysis data to capture the diversity of values for all variables across the atmospheric column, then, divides the atmospheric column into upper and lower levels along a model level that approximately separates the column into high-altitude and lower-altitude winds. We use model-level data, rather than pressure-level, to ensure that topography is adequately accounted for, and to prevent moisture loss due to abrupt changes in surface elevation (Keys et al. 2014). Previous comparisons of the WAM-2layers to models comprising many more atmospheric layers have found that the two-layer approach provides sufficient information for capturing the heterogeneity that can occur between upper and lower layers of the atmosphere (van der Ent et al., 2013; van der Ent & Tuinenburg, 2017). 1.5° spatial resolution reanalysis is sufficient to capture patterns in moisture transport and used for computational efficiency (Keys et al., 2014). For further detailed description of the WAM-2layers, refer to van der Ent et al. (2013, 2014).

We conduct backtracking analysis between 79.5°S and 79.5°N, and do not account for moisture from grid cells beyond these boundaries because of the potential for numerical errors with reduced grid cell size at

CLUETT ET AL.

10.1029/2020JD033543

4 of 15

Journal of Geophysical Research: Atmospheres

high latitudes. However, given the net northerly moisture flux to the Arctic in this region, the impact of this boundary is likely small (Figure S2; Text S1). We exclude the first and last years from analysis to minimize errors in backtracking, and report results as average monthly values for 1980–2013. The sum of backtracked evaporation from all grid cells represents total precipitation to the sink region for each month, and com-parison of tracked evaporation, ERA-interim precipitation, and instrumental precipitation records indicate that tracked evaporation robustly captures large-scale spatial and seasonal patterns in precipitation amount (Figure 1d).

2.3. Analysis

We classify grid cells as continental or oceanic based on the ERA-Interim land-sea-mask, with grid cells of greater or equal to 75% land classified as continental. We consider lakes and seas not connected directly to the ocean to be continental. We manually assign grid cells to the source regions distinguished in Fig-ures 2b–2d and shown in Figure S3. We calculate the average moisture transport distance as the tracked evaporation flux-weighted average distance from the middle of each source grid cell to the location of the meteorological/GNIP station within the sink region (Figure 1a). We calculate the evaporation flux-weighted average moisture source latitude.

2.4. Simulation of Precipitation Isotope Seasonality

We evaluate the sensitivity of precipitation δ2H seasonality to moisture source variations using a one-di-mensional Rayleigh distillation model forced by three scenarios of seasonal variations in the source-to-sink cooling of moisture to each domain, at monthly resolution (Table 1) (Fritz & Clark, 1997).

We calculate saturation vapor pressure (V) as a function of temperature (T, in Kelvin), according to Equa-tion 1 (Fritz & Clark, 1997).

3 2

0.0002 273.15 0.0111 273.15 0.321 273.15 4.8V T T T (1)

CLUETT ET AL.

10.1029/2020JD033543

5 of 15

Figure 2. Mean annual and monthly regional contributions of moisture sources to western Greenland domains. (a) Percentage of annual tracked evaporation from each grid cell. (b) Average total monthly regional contributions. Regional boundaries are illustrated in Figure S3. (c) Average fractional monthly regional contributions. (d) Average annual fractional regional contributions.

Journal of Geophysical Research: Atmospheres

Therefore, the fraction of original moisture (fm) remaining at any point along a distillation pathway is given by fm = V/V0, where V0 is the vapor pressure at the source. We calculate the resulting δ2H value of precipita-tion (δ2HPrecip) following Equation 2, where δ2HV0 is the starting vapor δ2H value and εl-v is the fractionation between vapor and liquid, a function of temperature in the sink region.

2 20 273.15 lnPrecip v l v l vH H fm T (2)

We calculate εl-v according to Equations  3 and  4, depending if the sink temperature is above freezing (T > 273.15 K) (Equation 3) or below freezing (T > 273.15 K) (Equation 4) (Majoube, 1971). We ignore the small cumulative impact of changing temperature on εl-v along the distillation pathway, and employ a con-stant value (273.15 K) (Fritz & Clark, 1997).

6 3

210 1024.844 76.248 52.612l v TT

(3)

6 3

210 1024.844 76.248 71.912i v TT

(4)

This model is not intended to represent event-scale processes, but to serve as a sensitivity test of the effects of seasonal changes in source temperature and location on final precipitation δ2H. Thus, we are primarily interested in the seasonal range of δ2HPrecip inferred in these sensitivity tests, rather than the absolute values.

In three scenarios of source-to-sink cooling (Table 1), we vary the sink temperature according to the av-erage seasonal cycle of the sink region as in the ERA-interim reanalysis data. For these three scenarios,

CLUETT ET AL.

10.1029/2020JD033543

6 of 15

# Sink temp Source temp δ2HV0 δ2HPrecip range

Comparison to OIPC δ2HPrecip Comparison to GNIP δ2HPrecip

Slope R RMSE Slope R RMSE

QAANAAQ

1 Seasonal Constant Constant 269‰ 0.47 0.98 7.7‰ 0.43 0.96 11‰

2 Seasonal Seasonal (no location ∆) Constant 124‰ 1.04 0.83 24‰ 1.00 0.85 22‰

3 Seasonal Seasonal + location ∆ Constant 180‰ 0.70 0.94 15‰ 0.65 0.94 14‰

4 Seasonal Seasonal + location ∆ Seasonal 193‰ 0.62 0.97 9.9‰ 0.57 0.96 11‰

GNIP — — 113‰ 1.04 0.98 8.1‰ — — —

OIPC — — 125‰ — — — 0.92 0.98 8.7‰

ILULISSAT

1 Seasonal Constant Constant 140‰ 0.59 0.91 13‰ — — —

2 Seasonal Seasonal (no location ∆) Constant 41‰ −0.58 −0.22 30‰ — — —

3 Seasonal Seasonal + location ∆ Constant 61‰ 1.21 0.68 22‰ — — —

4 Seasonal Seasonal + location ∆ Seasonal 74‰ 1.07 0.92 12‰ — — —

OIPC — — 88‰ — — — — — —

KANGILINNGUIT

1 Seasonal Constant Constant 72‰ 0.28 0.65 8.8‰ −0.01 −0.04 6.7‰

2 Seasonal Seasonal (no location ∆) Constant 60‰ −0.43 −0.69 8.4‰ 0.04 0.12 6.7‰

3 Seasonal Seasonal + location ∆ Constant 43‰ −0.58 −0.62 9.1‰ 0.07 0.13 6.7‰

4 Seasonal Seasonal + location ∆ Seasonal 38‰ 0.42 0.32 11‰ 0.16 0.21 6.6‰

GNIP — — 20‰ 0.92 0.54 5.7‰ — — —

OIPC — — 44‰ — — — 0.31 0.54 9.8‰

Table 1 Rayleigh Distillation Scenarios and Statistics on the Linear Regression of Monthly Simulated δ2HPrecip to Monthly Average δ2HPrecip as in OIPC and GNIP Data

Journal of Geophysical Research: Atmospheres

only source temperatures differ between these scenarios and we prescribe a starting vapor δ2H (δ2HV0) of −120‰, a reasonable estimate of mean annual source vapor (Good et al., 2015; Risi et al., 2012).

In the first scenario, we prescribe constant source temperatures as the average annual tracked evapora-tion-amount-weighted source temperature for each domain. These source temperatures represent the aver-age annual temperatures of the average annual moisture source of moisture to each sink region. In the sec-ond scenario, we calculate seasonally varying source temperatures assuming no change in source location throughout the year, but incorporating the monthly temperature variations of the average source region. We calculate these source temperatures as average monthly temperature of all contributing grid cells weighted by the average annual tracked evaporation flux. In the third scenario, we incorporate spatial variations in the source, calculating source temperatures as the average monthly evaporation-amount-weighted source temperature. The second and the third scenarios distinguish the impacts of variability in the seasonality of source temperatures and moisture source locations, to demonstrate the impact of changes in seasonal moisture source locations alone. Finally, in a fourth scenario, we introduce variations in the starting vapor isotopic composition according to source location, with temperatures as in the third scenario. We estimate average monthly δ2HV0 variations as the zonal averages at the average source latitude in the lowest level of the LMDZ isotope enabled general circulation model (Risi et al., 2012) (Figure S4).

To quantify the similarity between the Rayleigh distillation model scenarios and constraints on monthly δ2HPrecip values, we calculate the slope, Pearson correlation coefficient (R value) and root mean square error (RMSE) for the linear regression model between the monthly δ2HPrecip values of each scenario and monthly mean values from the OIPC and GNIP data (Table 1).

2.5. Simulation of Future Precipitation Isotope Seasonality

To evaluate the impact of projected moisture transport changes on future western Greenland δ2HPrecip sea-sonality (Bintanja & Selten, 2014; Singh et al., 2017), we simulate monthly δ2HPrecip for each domain under varying contributions from local and remote moisture sources. We simulate δ2HPrecip seasonality as in our fourth Rayleigh scenario, additionally prescribing incremental increases in moisture from local sources in the winter and remote sources in the summer (Figure 5c). We weight contributions of moisture from remote sources (North America, Atlantic Ocean south of 45°N, Pacific Ocean) and local sources (Northern North Atlantic, Labrador Sea, Baffin Bay, Canadian Arctic, and Arctic Ocean) by 150%–500%. These sensitivity tests isolate the impact of varying the fraction of local and remote moisture sources under modern condi-tions, but do not account for changes in source and sink temperatures which would likely be associated with these changes in moisture transport.

3. Results and Discussion3.1. Moisture Sources to Coastal Western Greenland

3.1.1. Annual Moisture Sources

The three analyzed western Greenland domains receive moisture locally from Baffin Bay or the Labrador Sea, and more remotely from North America, the Pacific, and the North Atlantic concentrated along the Gulf Stream (Figures  2a and  2d). Annual moisture contributions to Qaanaaq precipitation are approxi-mately evenly distributed among North American (22%), Atlantic (20%), Pacific (27%), and Arctic sources (23%). However, moving southward among domains, moisture from the Atlantic accounts for an increasing percentage of total annual precipitation. North American (23%), Pacific (19%), and Arctic (20%) sources contribute similar amounts of moisture to Ilulissat, while the Atlantic contributes a larger percentage (32%). To Kangilinnguit, the Atlantic contributes 43% of annual precipitation, predominating over North Ameri-can (24%), Pacific (15%), and Arctic (14%) sources. Moisture to Kangilinnguit is, on average, sourced from farther south (average source latitude = 42.4°N) than moisture to Ilulissat (45.5°N) and Qaanaaq (48.3°N), resulting from more southerly sources of local moisture and stronger contributions from the sub-tropical North Atlantic.

CLUETT ET AL.

10.1029/2020JD033543

7 of 15

Journal of Geophysical Research: Atmospheres

3.1.2. Seasonal Moisture Source Variations

We observe variations in monthly moisture sources to all domains, with similar broad patterns in the rela-tive importance of oceanic and continental moisture sources consistent with prior Greenland-scale analyses (Figures 2, 3, and S5–S7) (Nusbaumer et al., 2019; Sodemann et al., 2008a; Werner et al., 2001). However, spatial differences occur in the location and relative importance of the most local moisture sources, particu-larly during winter (Figure 2).

During summer (JJA), continentally sourced moisture from North America dominates all domains (Fig-ures 2b and 2c), increasing throughout spring (MAM) to a maximum of 42%–47% during July or August (Figure 3a), comparable to estimates of 45%–53% for the entire Greenland Ice Sheet (Nusbaumer et al., 2019) and of 42% at Summit (Werner et al., 2001). Throughout the summer, evapotranspiration from North Amer-ica strengthens and the region of strongest evapotranspiration migrates northward, increasing the average source latitude of continental moisture and reducing the distance over which continental moisture is trans-ported (Figures 3b and 3c). Oceanic contributions are least important during the summer (Figure 3a), when the sources of oceanic moisture are most remote (Figures 3b and 3c).

During winter (DJF), the majority of moisture is sourced from remote oceanic sources, making up 59%, 60%, and 69% of moisture to Qaanaaq, Ilulissat, and Kangilinnguit respectively (Figure 3a). For Ilulissat and Kangilinnguit, remote oceanic moisture is mainly sourced from the North Atlantic (42% and 54%, re-spectively). These values agree with studies for the entire Greenland Ice Sheet, which found 67% of winter moisture from the North Atlantic and 11%–14% of moisture from the Pacific (Nusbaumer et al., 2019), and are similar to estimates for Summit (Sodemann et al., 2008a; Werner et al., 2001). Strong moisture transport from the Gulf Stream during the winter to Kangilinnguit contrast with minimal moisture from this source to Qaanaaq, where remote oceanic moisture sources are partitioned between the Atlantic (30%) and Pacific (29%) and overall precipitation is low (Figure 2b). Local oceanic contributions (mainly from the Labrador

CLUETT ET AL.

10.1029/2020JD033543

8 of 15

Figure 3. Average seasonal cycles in moisture sources to western Greenland domains. (a) Fraction of tracked evaporation from continental (red) and oceanic (blue) sources. (b) Average distance between site of evaporation and site of precipitation for all tracked evaporation (black; error bars representing one interannual standard deviation) and average distance for continental (red) and oceanic (blue) components. (c) Average latitude of all tracked evaporation (black), and continental (red) and oceanic (blue) components.

Journal of Geophysical Research: Atmospheres

Sea and Baffin Bay) to all three domains reach relative and absolute maximums during fall (SON) and win-ter, peaking at 25%–35% of monthly precipitation between October and December. Whereas local moisture sources dominate in the winter at Kangilinnguit and Ilulissat, the presence of local sea ice prevents the same northward migration of the local moisture source to Qaanaaq (Figure S7). As a result, the most locally sourced oceanic moisture occurs in October for Qaanaaq (transport distance = 3,600 km), but January for Ilulissat (3,300 km) and Kangilinnguit (2,900 km). Continental sourced moisture to all domains is low and most remote through the fall and winter, when continental moisture contributions are limited to southeast-ern North America (Figures S5–S7).

3.2. Precipitation Isotope Seasonality Simulated From Source and Sink Temperature Variations

In the first Rayleigh Distillation scenario, we assume the source temperature constant at its annual average value; the seasonal cycle in the magnitude of cooling between the source and sink is determined by changes in the local sink temperature only. Therefore, the magnitude of cooling between the source and the sink is greater during winter than during summer for all sites (Figure 4a). As a result, precipitation occurs at a lower remaining fraction of moisture during winter than summer (Figure 4b), resulting in lower δ2HPrecip

CLUETT ET AL.

10.1029/2020JD033543

9 of 15

Figure 4. Simulated seasonal precipitation isotope patterns. (a) Average monthly temperatures at the site of precipitation (black; triangles down), mean tracked evaporation source (black; triangles up), monthly temperatures assuming constant moisture source (gray), and mean annual moisture source temperature (dashed line). (b) Fraction of moisture (fm) remaining in air mass at time of precipitation assuming variable source with seasonal temperature change (black), constant source with seasonal temperature change (gray), and constant source with constant temperature (dashed). (c) Simulated precipitation δ2H given the four scenarios, compared to observations of precipitation δ2H (solid blue; IAEA/WMO, 2015) and OIPC-modeled precipitation δ2H (dashed blue; Bowen, 2019).

Journal of Geophysical Research: Atmospheres

values during winter than summer (Figure 4c). Simulated δ2HPrecip seasonality is strongest for Qaanaaq with a range of 269‰, in comparison to 140‰ for Ilulissat and 72‰ for Kangilinnguit (Table 1). The observed difference in δ2HPrecip seasonality between Qaanaaq and Kangilinnguit could result from differences in local temperature seasonality alone. However, holding source temperature constant overestimates δ2HPrecip sea-sonality for both sites.

If we prescribe seasonal changes in the average moisture source temperature without invoking changes in source location as in the second Rayleigh Distillation scenario, lower winter source temperatures decrease cooling between the source and sink in the winter relative to the first scenario. Likewise, warmer source temperatures during summer increase cooling between the source and sink during the summer relative to the first scenario. Consequently, winter precipitation occurs at a slightly higher fraction of remaining mois-ture and summer precipitation occurs at a substantially lower fraction of remaining moisture than if a con-stant source temperature is assumed, resulting in dampened modeled δ2HPrecip seasonality of 124‰, 41‰, and 60‰ for Qaanaaq, Ilulissat, and Kangilinnguit respectively. In fact, assuming the seasonal variation of a constant annual source inverts the seasonal δ2HPrecip patterns simulated for Ilulissat and Kangilinnguit, yielding most negative δ2HPrecip values in the late summer and fall, and a negative slope when regressed against OIPC values.

If we vary source temperatures seasonally and according to source location as in the third Rayleigh Dis-tillation scenario, we find that seasonal variations in moisture source temperature are similar to the tem-peratures without accounting for location changes (Figure 3a). However, for all sites, warmer source tem-peratures occur during winter considering source location variations than if a constant source location is assumed. This difference is largest for Qaanaaq, where source temperature is warmer by 8.3°C during January in comparison to Ilulissat (∆4.5°C) and Kangilinnguit (∆2.5°C). Warmer winter source temper-atures increase δ2HPrecip seasonality for Qaanaaq (180‰) and Ilulissat (61‰) in comparison to the second scenario, while δ2H seasonality for Kangilinnguit (43‰) is reduced. Similar source temperatures among all sites are consistent with southward displacement of the most local source of moisture beyond the sea ice

CLUETT ET AL.

10.1029/2020JD033543

10 of 15

Figure 5. Simulated seasonal precipitation isotope patterns with prescribed increases in locally sourced winter moisture and remotely sourced summer moisture. (a) Fraction of modern moisture from remote (orange) and local (purple) sources, and fraction of moisture from local and remote sources given increases in local and remote moisture sources in the winter and summer, respectively, by 150%–500%. (b) Simulated modern precipitation δ2H as in the fourth Rayleigh simulation (black), and given prescribed changes in local (purple) and remote (orange) moisture sources. (c) Modern total monthly precipitation amount (black) and precipitation amounts given prescribed moisture increases.

Journal of Geophysical Research: Atmospheres

front, supported by comparatively long transport distances and similar average source latitudes for Qaa-naaq and Ilulissat during the winter and spring relative to Kangilinnguit (Figure 3). These results support that sea ice is an important factor in modulating the location and strength of local winter moisture sources (Nusbaumer et al., 2019), and promotes relatively low winter δ2HPrecip values in coastal regions near to and north of the seasonal sea ice front. Therefore, past or future increases in winter δ2HPrecip at sites north of the modern winter sea ice front could indicate reductions in local sea ice and strengthened local moisture contributions in response to reductions in local sea ice, amplifying the sensitivity of winter δ2HPrecip to local temperature change.

Varying source δ2HV0 induces additional δ2HPrecip seasonality for Qaanaaq and Ilulissat (Figure 4d). We vary the average monthly source δ2HV0 by 40‰–44‰ intra-annually among sites, with most negative values in the winter and most positive values in the summer (Figure S4). Incorporation of both seasonal temperature and δ2HV0 variability yields seasonal patterns and ranges in simulated δ2HPrecip similar to those observed for Kangilinnguit (38‰) and Qaanaaq (193‰), and an intermediate value for Ilulissat (74‰). The magnitude of seasonality which we simulate is greater than the observations for Qaanaaq, which could reflect greater sensitivity to transport conditions or kinetic fractionation processes at low temperatures which are not cap-tured in our simple model (Landais et al., 2012), or uncertainty in δ2HV0 for this northernmost site.

Additionally, an obvious signal to consider is the shift from predominantly continental to oceanic moisture sources between summer and winter. Deuterium-excess shows generally more negative values for both Qa-anaaq and Kangilinnguit during the summer than the winter (Figure S1), consistent with an increased pro-portion of recycled continentally sourced moisture during the summer (Wei & Lee, 2019). On a global scale, satellite-derived estimates of continental evapotranspiration are slightly 2H-depleted compared to oceanic evaporation, which could decrease summer δ2HPrecip at all sites (Good et al., 2015). However, isotope-ena-bled climate models demonstrate that transpiration yields more 2H-enriched atmospheric vapor than ocean evaporation, which would suggest continental evapotranspiration would increase summer δ2HPrecip at all sites (Risi et al., 2013). Although similarity in intra-annual patterns of continental moisture contributions to all sites precludes differences in continental and oceanic δ2HV0 from explaining differing δ2HPrecip season-ality between sites, better constraining how changes in continental moisture impact δ2HPrecip could improve our understanding of the controls on summer-biased δ2HPrecip proxy records.

Simulated intra-annual variability in δ2HPrecip highlights the importance of non-linearity in the relationship between local temperature and Rayleigh distillation. Following the Clausius-Clapeyron relationship, sen-sitivity of the fraction of moisture remaining, and thus the strength of distillation, to temperature changes increases at low temperatures (Fritz & Clark, 1997). In other words, temperature changes of the same mag-nitude result in larger changes in the fraction of moisture remaining at low temperatures than at warm tem-peratures. As a result of this asymmetry in temperature sensitivity, for precipitation at Kangilinnguit, which occurs under comparatively warm conditions, variations in source temperature due to moisture source location and temperature seasonality are sufficient to compensate for local temperature-driven δ2HPrecip seasonality. More dramatic variations in source temperature would be required to compensate for local tem-perature-driven δ2HPrecip seasonality at Qaanaaq, where low winter temperatures drive strong distillation. Following this logic, as regional temperatures increases, the resulting weaker distillation should increase the relative importance of moisture source variations on δ2HPrecip variability at all sites, even if moisture source and transport patterns do not change.

3.3. Reduced Precipitation Isotope Seasonality Under Future Moisture Transport Conditions

Anthropogenic warming is projected to increase future Arctic precipitation, with increases in the winter due to increased local evaporation and increases in the summer due to strengthened moisture transport from lower latitudes (Bintanja & Selten,  2014; Singh et  al.,  2017). In combination, our sensitivity tests indicate that the contrasting seasonality of these mechanisms will reduce δ2HPrecip seasonality at all sites (Figure 5b). The sensitivity of monthly δ2HPrecip to intensified contributions from local or remote moisture sources will depend on the impact of the increased moisture flux on the fractional contribution from the source (Figure 5a). Because remote moisture sources are already dominant during the summer, the influ-ence of increasing meridional transport in the summer is limited, even given enormous prescribed increases in remote moisture: amplifying the contributions of remote moisture sources by 500% results in summer

CLUETT ET AL.

10.1029/2020JD033543

11 of 15

Journal of Geophysical Research: Atmospheres

precipitation isotope values less than 10‰ more negative, an impact which would likely be canceled out by warmer local sink temperatures. On the other hand, modern winter precipitation to all sites is from more mixed local and remote sources, with approximately 40% of moisture from local sources and 60% of mois-ture from remote sources. Therefore, increasing local moisture contributions during the winter has a more pronounced impact on the fractional contributions from local and remote moisture sources and a great-er influence on δ2HPrecip, than increasing remote moisture contributions during the summer. Additionally, winter is associated with a stronger hemispheric temperature gradient than summer, and thus changing the proportion of local and remote moisture sources has a stronger influence on the average source temperature and resulting δ2HPrecip during winter than summer. While increases in local moisture during the summer would also promote enriched δ2HPrecip at all sites, we would expect the magnitude of change to be smaller than simulated for winter.

Although these sensitivity tests do not capture all changes that would impact precipitation isotope values in these regions in a warmer world, varying the contributions of remote and local moisture allows us to explore where and when precipitation isotopes are the most sensitive to changes in moisture source region. In addition to changes in the proportions of remotely and locally sourced moisture, projected moisture transport changes would likely be associated with warmer conditions at the source, sink, and throughout transport, and changes in δ2HV0. A more robust simulation of expected seasonal changes in δ2HPrecip would require constraint of seasonal changes in these parameters. Nevertheless, changing the proportion of re-mote and local moisture source contributions while assuming these other factors remain constant suggests that winter δ2HPrecip in coastal western Greenland is particularly sensitive to increases in local evaporation, with the strongest sensitivity for sites north of the modern seasonal sea ice front (Figure 5). Therefore, under warmer conditions, increases in local evaporation will likely promote more enriched δ2HPrecip than expected from increases in local temperatures alone.

4. ConclusionsWe demonstrate that moisture sources to coastal western Greenland exhibit strong intra-annual variations, and argue that patterns in moisture transport distance and cooling explain the observed meridional gradient in precipitation isotope seasonality. All sites receive moisture from a combination of oceanic and continen-tal sources, with an increasing predominance of Atlantic moisture moving southward. During the summer, moisture to all sites is primarily sourced from North America, whereas moisture from the northern North Atlantic and the Labrador Sea dominates in fall and winter. The Gulf Stream contributes particularly large portions of moisture to southern Greenland, so this region may be especially sensitive to changes in meridi-onal energy transport by this ocean current. The presence of seasonal sea ice in the Labrador Sea and Baffin Bay restricts the location of winter moisture sources, reducing local moisture sources north of the winter sea ice front, suggesting past movement of the sea ice front would influence winter precipitation isotope records from these regions.

During summer, more remotely sourced moisture to Kangilinnguit promotes lower δ2HPrecip values, whereas more locally sourced moisture during winter promotes higher δ2HPrecip values, compensating for the impact of local temperature seasonality. In comparison, moisture sources to Qaanaaq are most remote in winter and early spring, driving strong depletion in δ2HPrecip and strong seasonality. Dominant patterns in precipi-tation isotope seasonality to coastal western Greenland sites can be simulated in a simple Rayleigh model, illustrating the importance of seasonal source temperatures in dampening δ2HPrecip and non-linearity in the effect of local temperature on distillation, especially for winter precipitation at Qaanaaq.

An increase in precipitation in the Arctic has been identified as a robust feature of warming across simula-tions of the coming century, attributed to enhanced poleward moisture transport primarily in the summer and increased local evaporation during the fall and winter (Bintanja & Selten, 2014; Singh et al., 2017). For the three coastal western Greenland domains, increasing local moisture contributions during the win-ter has a more pronounced impact on δ2HPrecip than increasing remote moisture contributions during the summer. During the winter, increases in local evaporation promote more enriched δ2HPrecip, even if local temperatures are held constant. Therefore, increases in local temperatures will likely be compounded by moisture source changes in the winter, promoting more strongly enriched δ2HPrecip than would be expected

CLUETT ET AL.

10.1029/2020JD033543

12 of 15

Journal of Geophysical Research: Atmospheres

based on local temperature changes, particularly for sites north of the modern sea ice front. During the summer, increasing moisture transport from low latitudes could reduce summer δ2HPrecip and the sensitivity of summer δ2HPrecip to local temperature changes within each domain. However, the signal of the moisture source changes in summer δ2HPrecip will likely be limited because moisture is already sourced primarily from remote sources and the impact of the dynamic change in moisture source acts in opposition to local temper-ature. In combination, the projected moisture transport changes are likely to reduce seasonality in δ2HPrecip at the annual scale. The responsiveness of intra-annual patterns in δ2HPrecip to moisture source and transport controls suggests that coastal precipitation isotope records may be uniquely useful for understanding past and future precipitation dynamics and better understanding these processes during warm periods, particu-larly during the winter. However, further long-term monitoring of precipitation isotopes from these regions would be useful in understanding the isotopic response of future hydrological changes and interpreting past changes in these processes based on isotope proxy records.

Conflict of InterestThe authors declare no conflicts of interest relevant to this study.

Data Availability StatementData and code for analyses are archived on zenodo (https://doi.org/10.5281/zenodo.3709323).

ReferencesAxford, Y., de Vernal, A., & Osterberg, E. C. (2021). Past warmth and its impacts during the Holocene thermal maximum in Greenland.

Annual Review of Earth and Planetary Sciences, 49(1). https://doi.org/10.1146/annurev-earth-081420-063858Balascio, N. L., D'Andrea, W. J., Gjerde, M., & Bakke, J. (2018). Hydroclimate variability of high Arctic Svalbard during the Holocene in-

ferred from hydrogen isotopes of leaf waxes. Quaternary Science Reviews, 183, 177–187. https://doi.org/10.1016/j.quascirev.2016.11.036Berrisford, P., Dee, D., Poli, P., Brugge, R., Fielding, K., Fuentes, M., et al. (2011). The ERA-Interim archive, version 2.0. ERA report series.Bintanja, R., & Selten, F. M. (2014). Future increases in Arctic precipitation linked to local evaporation and sea-ice retreat. Nature,

509(7501), 479–482. https://doi.org/10.1038/nature13259Bonne, J.-L., Masson-Delmotte, V., Cattani, O., Delmotte, M., Risi, C., Sodemann, H., & Steen-Larsen, H. C. (2014). The isotopic composi-

tion of water vapour and precipitation in Ivittuut, southern Greenland. Atmospheric Chemistry and Physics, 14(9), 4419–4439. https://doi.org/10.5194/acpd-13-30521-201310.5194/acp-14-4419-2014

Bowen, G. J. (2008). Spatial analysis of the intra-annual variation of precipitation isotope ratios and its climatological corollaries. Journal of Geophysical Research, 113(D5). https://doi.org/10.1029/2007JD009295

Bowen, G. J. (2019). The online isotopes in precipitation calculator, version 3.1. Retrieved from http://www.waterisotopes.orgBowen, G. J., Cai, Z., Fiorella, R. P., & Putman, A. L. (2019). Isotopes in the water cycle: Regional- to global-scale patterns and applications.

Annual Review of Earth and Planetary Sciences, 47, 453–479. https://doi.org/10.1146/annurev-earth-053018-060220Bowen, G. J., Wassenaar, L. I., & Hobson, K. A. (2005). Global application of stable hydrogen and oxygen isotopes to wildlife forensics.

Oecologia, 143, 337–348. https://doi.org/10.1007/s00442-004-1813-yBuizert, C., Gkinis, V., Severinghaus, J. P., He, F., Lecavalier, B. S., Kindler, P., et al. (2014). Greenland temperature response to climate

forcing during the last deglaciation. Science, 345(6201), 1177–1180. https://doi.org/10.1126/science.1254961Ciais, P., & Jouzel, J. (1994). Deuterium and oxygen 18 in precipitation: Isotopic model, including mixed cloud processes. Journal of Geo-

physical Research, 99(D8), 16793–16803. https://doi.org/10.1029/94JD00412Corcoran, M. C., Thomas, E. K., & Morrill, C. (2021). Using a paired chironomid δ18O and aquatic leaf wax δ2H approach to recon-

struct seasonality on Western Greenland during the Holocene. Paleoceanogr Paleoclimatol, 36, e2020PA004169. https://doi.org/10.1029/2020PA004169

Cuffey, K. M., Clow, G. D., Alley, R. B., Stuiver, M., Waddington, E. D., & Saltus, R. W. (1995). Large arctic temperature change at the Wis-consin-Holocene glacial transition. Science, 270(5235), 455–458. https://doi.org/10.1126/science.270.5235.455

Dansgaard, W. (1964). Stable isotopes in precipitation. Tellus, 16(4), 436–468. https://doi.org/10.3402/tellusa.v16i4.8993Dufour, A., Zolina, O., & Gulev, S. K. (2016). Atmospheric moisture transport to the Arctic: Assessment of reanalyses and analysis of trans-

port components. Journal of Climate, 29(14), 5061–5081. https://doi.org/10.1175/JCLI-D-15-0559.1Fritz, P., & Clark, I. D. (1997). Environmental isotopes in hydrogeology. CRC press. https://doi.org/10.4095/208891Good, S. P., Noone, D., Kurita, N., Benetti, M., & Bowen, G. J. (2015). D/H isotope ratios in the global hydrologic cycle. Geophysical Research

Letters, 42(12), 5042–5050. https://doi.org/10.1002/2015GL064117Hendricks, M. B., DePaolo, D. J., & Cohen, R. C. (2000). Space and time variation of δ18O and δD in precipitation: Can paleotemperature be

estimated from ice cores? Global Biogeochemical Cycles, 14(3), 851–861. https://doi.org/10.1029/1999GB001198IAEA/WMO (2015). Global Network of Isotopes in Precipitation. The GNIP Database. Retrieved from https://nucleus.iaea.org/wiserJohnsen, S. J., Dahl-Jensen, D., Gundestrup, N., Steffensen, J. P., Clausen, H. B., Miller, H., et al. (2001). Oxygen isotope and palaeotemper-

ature records from six Greenland ice-core stations: Camp Century, Dye-3, GRIP, GISP2, Renland and NorthGRIP. Journal of Quaternary Science, 16(4), 299–307. https://doi.org/10.1002/jqs.622

Johnsen, S. J., Dansgaard, W., & White, J. W. C. (1989). The origin of Arctic precipitation under present and glacial conditions. Tellus B: Chemical and Physical Meteorology, 41(4), 452–468. https://doi.org/10.3402/tellusb.v41i4.15100

CLUETT ET AL.

10.1029/2020JD033543

13 of 15

AcknowledgmentsThis work is supported by an NSF Graduate Research Fellowship to AAC (No. 1645677) and NSF grants 1652274 and 1737716 to EKT. The authors are grateful for constructive comments from three reviewers.

Journal of Geophysical Research: Atmospheres

Jouzel, J., Alley, R. B., Cuffey, K. M., Dansgaard, W., Grootes, P., Hoffmann, G., et al. (1997). Validity of the temperature reconstruction from water isotopes in ice cores. Journal of Geophysical Research, 102(C12), 26471–26487. https://doi.org/10.1029/97JC01283

Kattsov, V. M., & Walsh, J. E. (2000). Twentieth-century trends of Arctic precipitation from observational data and a climate model simula-tion. Journal of Climate, 13(8), 1362–1370. https://doi.org/10.1175/1520-0442(2000)013<1362:tctoap>2.0.co;2

Keys, P. W., Barnes, E. A., Van Der Ent, R. J., & Gordon, L. J. (2014). Variability of moisture recycling using a precipitationshed framework. Hydrology and Earth System Sciences, 18(10), 3937–3950. https://doi.org/10.5194/hess-18-3937-2014

Keys, P. W., van der Ent, R. J., Gordon, L. J., Hoff, H., Nikoli, R., & Savenije, H. H. G. (2012). Analyzing precipitationsheds to understand the vulnerability of rainfall dependent regions. Biogeosciences, 9(2), 733–746. https://doi.org/10.5194/bg-9-733-2012

Klein, E. S., Cherry, J. E., Young, J., Noone, D., Leffler, A. J., & Welker, J. M. (2015). Arctic cyclone water vapor isotopes support past sea ice retreat recorded in Greenland ice. Scientific Reports, 5(1), 1–9. https://doi.org/10.1038/srep10295

Klein, E. S., & Welker, J. M. (2016). Influence of sea ice on ocean water vapor isotopes and Greenland ice core records. Geophysical Research Letters, 43(24). https://doi.org/10.1002/2016GL071748

Landais, A., Ekaykin, A., Barkan, E., Winkler, R., & Luz, B. (2012). Seasonal variations of 17 O-excess and d-excess in snow precipitation at Vostok station, East Antarctica. Journal of Glaciology, 58, 725–733. https://doi.org/10.3189/2012JoG11J237

Lasher, G. E., Axford, Y., McFarlin, J. M., Kelly, M. A., Osterberg, E. C., & Berkelhammer, M. B. (2017). Holocene temperatures and isotopes of precipitation in Northwest Greenland recorded in lacustrine organic materials. Quaternary Science Reviews, 170, 45–55. https://doi.org/10.1016/j.quascirev.2017.06.016

Leng, M. J., & Marshall, J. D. (2004). Palaeoclimate interpretation of stable isotope data from lake sediment archives. Quaternary Science Reviews, 23(7), 811–831. https://doi.org/10.1016/j.quascirev.2003.06.012

Levin, N. E., Zipser, E. J., & Cerling, T. E. (2009). Isotopic composition of waters from Ethiopia and Kenya: Insights into moisture sources for eastern Africa. Journal of Geophysical Research, 114(D23). https://doi.org/10.1029/2009JD012166

Li, L., & Garzione, C. N. (2017). Spatial distribution and controlling factors of stable isotopes in meteoric waters on the Tibetan Pla-teau: Implications for paleoelevation reconstruction. Earth and Planetary Science Letters, 460, 302–314. https://doi.org/10.1016/j.epsl.2016.11.046

Lindborg, T., Rydberg, J., Tröjbom, M., Berglund, S., Johansson, E., Löfgren, A., et al. (2016). Biogeochemical data from terrestrial and aquatic ecosystems in a periglacial catchment, West Greenland. Earth System Science Data, 8(2), 439–459. https://doi.org/10.5194/essd-8-439-2016

Linderholm, H. W., Nicolle, M., Francus, P., Gajewski, K., Helama, S., Korhola, A., et al. (2018). Arctic hydroclimate variability during the last 2000 years: Current understanding and research challenges. Climate of the Past, 14, 473–514. https://doi.org/10.5194/cp-14-473-2018

Majoube, M. (1971). Fractionnement en oxygène 18 et en deutérium entre l'eau et sa vapeur. Journal de Chimie Physique, 68, 1423–1436. https://doi.org/10.1051/jcp/1971681423

Mernild, S. H., Hanna, E., McConnell, J. R., Sigl, M., Beckerman, A. P., Yde, J. C., et al. (2015). Greenland precipitation trends in a long-term instrumental climate context (1890-2012): Evaluation of coastal and ice core records. International Journal of Climatology, 35(2), 303–320. https://doi.org/10.1002/joc.3986

NOAA (2020). The global historical climatology network. National Center for Environmental Information. Retrieved from https://www.ncdc.noaa.gov/ghcn-monthly

Nusbaumer, J., Alexander, P. M., LeGrande, A. N., & Tedesco, M. (2019). Spatial shift of Greenland moisture sources related to enhanced Arctic warming. Geophysical Research Letters, 46(24), 14723–14731. https://doi.org/10.1029/2019GL084633

Putman, A. L., Feng, X., Sonder, L. J., & Posmentier, E. S. (2017). Annual variation in event-scale precipitation δ2H at Barrow, AK, reflects vapor source region. Atmospheric Chemistry and Physics, 17(7), 4627–4639. https://doi.org/10.5194/acp-17-4627-2017

Rawlins, M. A., Steele, M., Holland, M. M., Adam, J. C., Cherry, J. E., Francis, J. A., et al. (2010). Analysis of the Arctic system for freshwa-ter cycle intensification: Observations and expectations. Journal of Climate, 23(21), 5715–5737. https://doi.org/10.1175/2010jcli3421.1

Risi, C., Noone, D., Frankenberg, C., & Worden, J. (2013). Role of continental recycling in intraseasonal variations of continental moisture as deduced from model simulations and water vapor isotopic measurements. Water Resources Research, 49(7), 4136–4156. https://doi.org/10.1002/wrcr.20312

Risi, C., Noone, D., Worden, J., Frankenberg, C., Stiller, G., Kiefer, M., et al. (2012). Process-evaluation of tropospheric humidity simulated by general circulation models using water vapor isotopic observations: 2. Using isotopic diagnostics to understand the mid and upper tropospheric moist bias in the tropics and subtropics. Journal of Geophysical Research, 117(D5). https://doi.org/10.1029/2011JD016623

Rozanski, K., Araguás-Araguás, L., & Gonfiantini, R. (1993). Isotopic Patterns in Modern Global Precipitation. In P. K. Swart, K. C. Lohmann, J. Mckenzie, & S. Savin (Eds.), Climate change in continental isotopic records (pp. 1–36). American Geophysical Union. https://doi.org/10.1029/GM078p0001

Severinghaus, J. P., Sowers, T., Brook, E. J., Alley, R. B., & Bender, M. L. (1998). Timing of abrupt climate change at the end of the Younger Dryas interval from thermally fractionated gases in polar ice. Nature, 391, 141–146. https://doi.org/10.1038/34346

Singh, H. K. A., Bitz, C. M., Donohoe, A., & Rasch, P. J. (2017). A source-receptor perspective on the polar hydrologic cycle: Sources, sea-sonality, and Arctic-Antarctic parity in the hydrologic cycle response to CO2 doubling. Journal of Climate, 30(24), 9999–10017. https://doi.org/10.1175/JCLI-D-16-0917.1

Sodemann, H., Masson-Delmotte, V., Schwierz, C., Vinther, B. M., & Wernli, H. (2008b). Interannual variability of Greenland winter pre-cipitation sources: 2. Effects of North Atlantic Oscillation variability on stable isotopes in precipitation. Journal of Geophysical Research, 113(D12). https://doi.org/10.1029/2007JD009416

Sodemann, H., Schwierz, C., & Wernli, H. (2008a). Interannual variability of Greenland winter precipitation sources: Lagrangian moisture diagnostic and North Atlantic Oscillation influence. Journal of Geophysical Research, 113(D3). https://doi.org/10.1029/2007JD008503

Thomas, E. K., Briner, J. P., Ryan-Henry, J. J., & Huang, Y. (2016). A major increase in winter snowfall during the middle Holocene on western Greenland caused by reduced sea ice in Baffin Bay and the Labrador Sea. Geophysical Research Letters, 43(10), 5302–5308. https://doi.org/10.1002/2016GL068513

Thomas, E. K., Castañeda, I. S., McKay, N. P., Briner, J. P., Salacup, J. M., Nguyen, K. Q., & Schweinsberg, A. D. (2018). A wetter Arctic coin-cident with hemispheric warming 8,000 years ago. Geophysical Research Letters, 45(19), 637–710. https://doi.org/10.1029/2018GL079517

Thomas, E. K., Hollister, K. V., Cluett, A. A., & Corcoran, M. C. (2020). Reconstructing arctic precipitation seasonality using aquatic leaf wax δ2H in lakes with contrasting residence times. Paleoceanography and Paleoclimatology, 35(7). https://doi.org/10.1029/2020PA003886

van der Ent, R. J., & Savenije, H. H. G. (2013). Oceanic sources of continental precipitation and the correlation with sea surface tempera-ture. Water Resources Research, 49(7), 3993–4004. https://doi.org/10.1002/wrcr.20296

van der Ent, R. J., Savenije, H. H. G., Schaefli, B., & Steele-Dunne, S. C. (2010). Origin and fate of atmospheric moisture over continents. Water Resources Research, 46(9). https://doi.org/10.1029/2010WR009127

CLUETT ET AL.

10.1029/2020JD033543

14 of 15

Journal of Geophysical Research: Atmospheres

van der Ent, R. J., & Tuinenburg, O. A. (2017). The residence time of water in the atmosphere revisited. Hydrology and Earth System Scienc-es, 21, 779–790. https://doi.org/10.5194/hess-21-779-2017

van der Ent, R. J., Wang-Erlandsson, L., Keys, P. W., & Savenije, H. H. G. (2014). Contrasting roles of interception and transpiration in the hydrological cycle - Part 2: Moisture recycling. Earth System Dynamics, 5(2), 281–326. https://doi.org/10.5194/esdd-5-281-2014

Wei, Z., & Lee, X. (2019). The utility of near-surface water vapor deuterium excess as an indicator of atmospheric moisture source. Journal of Hydrology, 577, 123923. https://doi.org/10.1016/j.jhydrol.2019.123923

Werner, M., Heimann, M., & Hoffmann, G. (2001). Isotopic composition and origin of polar precipitation in present and glacial climate simulations. Tellus B: Chemical and Physical Meteorology, 53(1), 53–71. https://doi.org/10.3402/tellusb.v53i1.16539

Reference From the Supporting InformationCluett, A. A., & Thomas, E. K. (2020). Resolving the combined influences of inflow and evaporation on western Greenland lake water

isotopes to inform paleoclimate interpretations. Journal of Paleolimnology, 1–18.

CLUETT ET AL.

10.1029/2020JD033543

15 of 15