Skip to main content
Dryad logo

Chemical effects of snowmelt on an alpine lake in the Wind River Range, WY


Ganz, Taylor; Benoit, Gaboury (2021), Chemical effects of snowmelt on an alpine lake in the Wind River Range, WY, Dryad, Dataset,


Nitrogen deposition from air pollution is increasingly reaching alpine lakes where the addition of nitrate and ammonium to sensitive surface waters can cause acidification and or eutrophication. Thirty years of sampling in the Wind River Range, WY have shown some lakes increasing in nitrogen. We sought to determine (1) if nutrient concentrations in Deep Lake increase during snowmelt when atmospheric deposition is released from the snowpack and (2) assess if the sampling season, location, meteorological factors, and time of day samples are collected influence lake chemistry metrics, to inform monitoring. We analyzed water samples from the outlet of Deep Lake in peak snowmelt (June) and from the inlet, outlet, and middle of Deep Lake when the basin was snow free (August). In June, outlet samples were more acidic, and nitrogen content was three times August levels. Acid neutralizing capacity (ANC) declined with snowmelt. August inlet samples were higher in nutrients than outlet and mid-lake samples. Our results indicate that atmospheric pollution in the snowpack enters the lake with snowmelt. Although Deep Lake has not acidified, ANC levels indicate a risk of episodic acidification if nitrogen deposition continues to increase. When monitoring lakes at risk for episodic acidification, sampling during the late snowmelt pulse should be prioritized. Simplified sampling protocols may be used in some lakes, as epilimnion and outlet samples were nearly identical. The time of day and cloud cover did not affect lake chemistry, while wind speed and precipitation weakly increased August ANC and June pH, respectively.


The Wind River Range extends roughly 250 kilometers, forming the continental divide through west-central Wyoming as part of the Rocky Mountains and the southeastern arm of the Greater Yellowstone Ecosystem (GYE). Rugged, glacially sculpted terrain, a high density of alpine lakes (over 1,500 lakes across 173,374 ha; Grenon et al. 2010; McMurray et al. 2013), cold temperatures, thin or absent soils, and a short growing season (less than 40 consecutive nights above freezing per year) characterize the Wind River Range ( Archean granitic rock, granitic gneiss, and migmatites are predominant across the range (Frost et al. 1998). Precipitation comes primarily in the form of snow and averages 100-130 cm total annual water equivalent at the lower-to-mid elevations and reaches up to 150 cm or more along the continental divide (Hall et al. 2012). An estimated two-thirds of the annual precipitation runs off as stream flow, with the rest attributed to evapotranspiration (Foster and Hall 1981; Hall et al. 2012). While the exact proportion of annual water derived from snow in the Wind River Range is undocumented, snowfall, snow cover, and the timing of snowmelt have been identified as the primary drivers of stream discharge for the region, emphasizing the dominance of snow in these hydrologic systems (Hall et al. 2012, 2015). Snowmelt from the western slopes of the Wind River Range drains to the Green River, providing water to the largest tributary of the Colorado River.

Two oil and gas extraction sites, the Pinedale Anticline and Jonah Project Area (Fig. 1), are located 40 km west and upwind of the southwestern Wind River Range in nearby Boulder, WY, and have been identified as a source of reactive nitrogen to the Southern Wind River Range, contributing the highest volume of nitrogen pollution in winter months (Zhang et al. 2018). The Jim Bridger coal plant 110 km to the south may provide an additional local source of reactive nitrogen to the mountain range. These extraction sites are sources of nitrogen in the form of nitrate, but they may also be a source of ammonia to the Wind River Range (Brahney et al. 2015; McMurray et al. 2013). Additional sources of nitrogen to the GYE also come from regional and long-distance transport from agriculture, fire and other sources (Zhang et al. 2018).

Deep Lake (elevation 3,218 m; 42.719084°N, 109.172130°W; Fig. 1) basin has a 190 ha watershed that lies in the southwestern Wind River Range within the Bridger Wilderness and is estimated to be 90% barren 7% grass, and 3% forest (Brahney 2012). Parent material in the basin is predominantly granitic, which breaks down slowly and thus has limited contribution of base cations to the total ANC (Brahney 2012; Mast et al. 1990, 2001; Turk and Spahr 1991). Like many high alpine lakes in the Wind River Range, the inlet and lake surface remain frozen through June, and water enters the lake diffusely from all sides as snow melts. In August, water primarily enters the lake through one main inlet, though groundwater contribution is possible but unknown (Fig. 1). Deep Lake is 24.5 ha in area and 27.0 m deep (Grenon et al. 2010). Deep Lake is not fed by any glaciers or permanent snowfields and all inputs thus reflect annual snowmelt, current precipitation, or possibly ground water which we were unable to examine. Given the watershed area, rates of evapotranspiration, and assuming 110 cm total annual water equivalence, mean annual runoff is estimated to be 1.4 × 106 m3 (Foster and Hall 1981; Hall et al. 2012).

Outflow from the lake can reflect the chemistry of the epilimnion when the lake is stratified, or a mix of layers following lake turnover (Hutchinson 1975). Thus, water chemistry at the outlet can be a function of input occurring over multiple years. Assuming the average depth is half the maximum depth of the lake, lake volume is calculated to be 3.3 × 106 m3 and mean residence time in Deep Lake is estimated to be 2.3 years, by equation (1) (Winchester 1968; Hutchinson 1975).

                        Water residence time (yrs) = Lake Volume (m3) / Mean Annual Runoff (m3/yr)    (1)

Detailed bathymetry of the lake is not available to form a more accurate estimate. We selected a value of one half because of the steepness of the surrounding terrain (Fig. 1). Our approximation is probably an upper limit to the volume of the lake, and consequently, the water residence time. It is important to note that traditional thinking about water residence time can be somewhat misleading for a water body like Deep Lake. In this system, almost all of the flow occurs during the short period of snowmelt. If this discharge continued throughout the year, water residence time would be much shorter. Instead, for most of the year, discharge is close to zero and the water in the lake through the colder months has a much longer effective residence time. Though it was not observed in the course of our study, in many years the outlet of Deep Lake dries up from late summer through autumn. Taken together, this means that all of the lake water that we measured (outlet and mid-lake) had an effective water residence time that was much shorter than the average residence time of all water in the lake. It is possible to estimate the magnitude of these effects. When the lake is stratified, we measured the epilimnion to be 9 m thick, and water residence time for this layer alone is 1.4 years. Furthermore, if most of the flow through the lake occurs in three of the twelve months, then this number reduces to an effective water residence time during the time we sampled of about four months.

Snow depth, snow-water equivalent and precipitation data referenced in this analysis were from the National Resources Conservation Service Big Sandy Opening snow-telemetry (SNOTEL) site (elevation 2770 m; 42.64580°N, 109.25965°W) located 10.6 km southwest of Deep Lake (National Water and Climate Center 2016). Annual precipitation at this site was reported to be 107% of average by the June sampling period and 97% of average by the August sampling period (National Water and Climate Center 2016).

Study Design

Sampling of Deep Lake occurred during two time intervals in the summer of 2016: (1) peak snowmelt (June 1-18) and, (2) peak primary production (August 7-13). In June, we collected grab samples of water at the outlet twice per day, in the morning (7:00-9:00) and evening (17:00 – 19:00) for 19 days along with one cycle of samples collected at the outlet every hour for 24 hours (8:00 June 9, 2016 to 8:00 June 10, 2016). We were unable to collect water samples from the mid-lake surface and lake inlet in June because those sites were frozen. In August, water samples were collected at the outlet, mid-lake, and inlet twice per day for 7 days, in the morning (7:00-10:00) and in the evening (17:00-19:00) as weather conditions allowed. A 24-hour sampling cycle was not conducted in August.

All samples were collected from the epilimnion at 10 cm depth and filtered with 0.45 μm nominal pore size Polyethersulfone Sterile Syringe Filters (N.A. PN 28145-505) VWR® attached to BD 60 ml Luer-LokTM syringes (REF 309653) VWR®. We collected 100 ml of filtered water via two syringe pulls, and stored the water in 125 ml polyethylene bottles, allowing 25 ml of headspace in the bottle to prevent rupture when they were frozen for long term storage. Neither acid nor other preservatives were added to the bottles. Bottles and syringes were rinsed with lake water (collected about 3m downstream of the sampling location) and syringes were not reused after sample collection. Samples were kept at approximately 1-3 ºC while in the field by submersing them in meltwater downstream of the sampling site. The bottles were immediately frozen following transport from the field until they were thawed for analysis. Water temperature and pH were measured at each sample collection using an Oakton® pH 150 Waterproof Portable pH/mV/Temperature Meter (EW-35614-90).

For every sampling event we observed meteorologic conditions, recording wind speed, and proportion of cloud cover following the Bridger-Teton National Forest Wind River Mountains Air Quality Monitoring Program Methods Manual (U.S. Department of Agriculture, Forest Service 2002). We also qualitatively recorded any precipitation since the last sampling period noting no precipitation as 0, precipitation since the last sampling event as 1, and precipitation during the sampling event as 2.

A Solinst® Levelogger® Jr. Edge M5 (110241) was submerged 0.4 km downstream of the lake outlet and recorded pressure every 5 minutes for the duration of the sampling periods. An adjacent, terrestrial Solinst® Levelogger® Edge M5 (110023) simultaneously recorded changes in atmospheric pressure. Using the difference between the two devices, we calculated downstream water depth at 5 minute intervals, corrected for variation in barometric pressure.

In August we recorded temperature and pH every meter to 19-m (the maximum reach of our measuring equipment) at mid-lake to evaluate depth profiles and identify a thermocline with a YSI® 556 Handheld Multi Parameter Instrument (556-01). The thermocline is a boundary between the epilimnion and hypolimnion, which limits mixing between the layers and allows flow to occur within the epilimnion (Lerman et al. 1995). An inflection point in the temperature gradient is used to identify the epilimnion, and a lake is considered strongly stratified when there is a temperature gradient of > 4°C between the surface and 60% of its depth (16.0 m for Deep Lake) (Landers et al. 1987; Lerman et al. 1995).

Chemical Analysis

We used an ICP-OES Optima 3000 (Perkin Elmer, Waltham, MA) to measure concentrations of strong base cations (Ca2+, K+, Mg2+ and Na+), an Ion Chromatograph (IC) DX500 (Dionex, Sunnyvale, CA), to measure strong acid anions (NO3-, SO4- and Cl-), and an Astoria 2 Flow Analyzer (Astoria Pacific, Clackamas, OR) to measure ammonium, total nitrogen, and total phosphorus. ANC was determined by gran titration. Duplicates were collected in the field for one-tenth of all samples, and ten percent of collected water samples were split and run as lab replicates for QA/QC purposes. Blanks of deionized water were stored in sample collection bottles, frozen with field samples for storage, and also analyzed for QA/QC, but not transported into the field. Strong base cation concentrations were calculated as the sum of the normal (eq/L) concentrations of Ca2+, K+, Mg2+ and Na+. The strong acid anion concentrations were taken to be the sum of the normal (eq/L) concentrations of NO3-, Cl-, and SO42-.

The ratio of nitrogen to phosphorus was calculated from the molarity of each sample as an indicator of nutrient limitation in the lake. In an analysis 2053 lakes and nearly 90 nitrogen and phosphorus lake enrichment experiments from Norway, Sweden and the Rocky Mountains, USA, Elser et al. (2009) found that lakes in low nitrogen-deposition areas had molar total nitrogen to total phosphorus ratio below 44.2 and were generally nitrogen-limited whereas lakes above a total nitrogen to total phosphorus ratio of 110 were consistently in high deposition areas and limited by phosphorus. Similarly, a synthesis of 221 lakes from 14 countries found that nitrogen-limitation was consistent when the total nitrogen to total phosphorus ratio was below ~31 and total phosphorus was above 0.003mg P/L (Downing and McCauley 1992). Dissolved inorganic nitrogen to total phosphorus ratios have been suggested as a better indicator of nutrient limits on phytoplankton growth, and by this metric nitrogen-limitation transitions to phosphorus-limitation when molar dissolved inorganic nitrogen to total phosphorus ratios shift from 3.3 to 7.5, with higher ratios typical of alpine lakes with moderate levels of nitrogen-deposition (Bergström 2010).

We focused our analysis on testing for changes in concentrations of total nitrogen, total phosphorus, titrated ANC, and the sums of the strong base cations and strong acid anions. To assess differences in these lake chemistry metrics, we ran ANOVAs testing the effect of sampling season (outlet samples only), sampling location (August only), and time of day the sample was collected. We used simple linear regression to determine if ion concentrations became more dilute during June sampling interval at the lake outlet over time. In lake monitoring programs, researchers often wonder if meteorological factors influence the chemistry of epilimnion samples, and thus should be accounted for in sampling protocols. We used general linear regression to test the influence of cloud cover, precipitation, and wind speed on major lake chemistry parameters in both June and August. We took p < 0.05 as our threshold for significance for all analyses.

Reported uncertainties for summary values by location, season, and time of day are given as the standard deviation. Uncertainties for measurements of Ca2+, K+, Mg2+, Na+, NO3-, SO4-, Cl-, NH4+, total nitrogen, total phosphorus, and titrated ANC were taken as the square root of the mean of the squares of the difference in replicates for each measure. Uncertainty values for the sum of the strong base cations and strong acid anions were taken to be the square root of the sum of the squared uncertainties of their components.

Usage Notes

Units and notes on detection limits are included in the header of the master file.


Institute for Biospheric Studies, Yale University

Yale School of Forestry and Environmental Studies

Ucross High Plains Stewardship Initiative

Ucross High Plains Stewardship Initiative