Groundwater attenuation of summer stream temperatures favors deeper intrusion depths into Lake George, NY

Temperature is one of the most important factors to consider when evaluating how streams function because of its influence on physical, chemical, and biological processes. Temperature is negatively related to water density and viscosity, which can impact stream intrusion depth into a lake (Laborde et al. 2010; Cortés et al. 2014), sediment infiltration rates (Constantz & Murphy 1991), and stream discharge (Constantz et al. 1994). Warmer water reduces gas solubility while enhancing biological oxygen demand (BOD), which can mobilize phosphorus (Liikanen et al. 2002) and trace metals from sediment in reduced conditions (Von Gunten et al. 1991). Since nearly all aquatic organisms are ectothermic, temperature influences metabolic rates similarly throughout trophic levels (Gillooly et al. 2001) with enzymatic activity approximately doubling with each 10 °C rise in temperature (Black 2012).

Many streams derive the majority of their discharge from groundwater and thus the headwater temperatures are similar to groundwater with water temperature trending toward air temperature with downstream flow (Sullivan et al. 1990). The extent of the temperature change depends on climate, riparian vegetation, stream morphology, and groundwater inputs (Sullivan & Adams 1991). Climate often is the dominating factor influencing water temperature with absorption of solar radiation being the primary source of warming (Morin & Couillard 1990; Webb & Zhang 1997; Evans et al. 1998; Johnson 2003). Riparian vegetation can insulate small streams by shading the water and the adjacent landscape (Sweeney 1993; Dong et al. 1998; Johnson 2004); however, the insulating effect of shading diminishes as stream width increases (Poole & Berman 2001). Stream morphology changes the surface area to volume ratio and can promote the exchange of stream water with the hyporheic zone (Brunke & Gonser 1997; Poole & Berman 2001; Webb et al. 2008). Hyporheic water temperatures generally are warmer than stream temperatures during the winter and cooler during the summer because a portion of the water originates from groundwater, which maintains a consistent year-round temperature that approximates the mean annual air temperature (Brunke & Gonser 1997; Hayashi & Rosenberry 2002). Streams heavily influenced by groundwater inputs exhibit attenuated year-round temperatures (Holmes 2000).

Ultimately, it is the combination of these factors that determine stream temperature, and the temperature difference between a stream and its receiving waterbody will dictate the stream intrusion depth (Fischer et al. 1979; Killworth & Carmack 1979). When stream inputs are warmer and less dense than the lake, they intrude as overflow. When stream inputs are cooler and thus denser, they intrude as underflow until a level of neutral buoyancy is reached (Alavian et al. 1992). In a well-mixed lake, all inputs will be mixed regardless of intrusion depth. During periods of stratification, however, intrusion depth will dictate at what depth in the water column stream constituents become available (MacIntyre et al. 2006; Cortés et al. 2014).

Intrusion depth in Lake George is of great interest because it may impact the regular occurrence of hypolimnetic oxygen depletion in the Caldwell Sub-basin, the southernmost sub-basin of the lake. The land surrounding the Caldwell Sub-basin is the most developed area in the Lake George watershed, based on property tax records from the eight towns and villages within the Lake George watershed; with roughly half of the ∼9,000 total buildings in the Lake George watershed located in this area. Greater urbanization coupled with the south to north flow of the lake has resulted in a consistent phosphorus and chlorophyll gradient in the lake that decreases as water flows north (Boylen et al. 2014). The higher nutrient concentration in the Caldwell Sub-basin and subsequent increased phytoplankton biomass has been indicated as a contributor to hypolimnetic oxygen depletion (Boylen et al. 2014). The greater organic biomass reaching the sediment surface enhances the BOD, lowering the hypolimnetic oxygen concentration during stratification with the lowest oxygen concentrations observed in October and November just before fall turnover. Oxygen depletion in Lake George becomes a concern when dissolved oxygen (DO) concentrations fall below 4 mg/L because at ∼10 °C, which is roughly the hypolimnetic temperature during early fall, 4 mg/L dissolved oxygen is equivalent to 35% saturation. Fish sensitive to low dissolved oxygen will avoid such areas (Arend et al. 2011) and oxygen concentrations ≤30% saturation result in the release of sediment-bound phosphorus (Eichler & Boylen 2011). In addition, for lakes classified AA-Special in New York State, including Lake George, ‘at no time shall the DO concentration be less than 4.0 mg/L’ (6 CRR-NY 703.3). Minimum hypolimnetic oxygen concentrations in the Caldwell Sub-basin were ≥4 mg/L during five years between 1980 and 2010 and for three of those years the minimum oxygen concentration was 4 mg/L.

The goal of this research was to determine: (1) if thermal regimes vary among the major streams entering Lake George, and if so, identify the factors that may be primarily responsible for the observed differences; (2) estimate the intrusion depth of each stream based on density; and (3) identify possible impacts of varying intrusion depths to the hypolimnetic oxygen depletion in Lake George. A better understanding of intrusion depth, entrainment, and mixing during stratification has broader implications on nutrient availability, primary production, and food web dynamics that could be applied to all waterbodies.

The lake has chemically distinct South and North basins that are separated by a shallow sill in the Narrows, ∼18 km from the south end. The South Basin, which is the focus of this study, consists of two sub-basins, Caldwell and Dome Island. The headwater and most southern sub-basin, Caldwell, has a maximum depth of 31 m and receives discharge from East, West, and English Brook (Figure 1). The Dome Island Sub-basin is the largest sub-basin with a maximum depth of ∼60 m (Boylen & Kuliopulos 1981) and receives discharge from Finkle, Indian, and Shelving Rock Brook. These six brooks are a sub-set of the 10 largest tributaries entering Lake George and drain 23% of the entire watershed (Swinton et al. 2015). The level of development ranges from the most impacted in East and West Brook watersheds to no development in the Shelving Rock Brook watershed. All the tributaries are relatively narrow with tree canopy covering the majority of the stream length.

During the last glaciation, the Laurentian Ice sheet stripped away most of the soil cover from higher elevations and deposited a greater proportion of sand in the southern end of the watershed with finer clays deposited in the North Basin. Upland deposits of sands and gravels are sporadic and infrequent with varying thickness overlaying fractured bedrock. Lower elevations have more consistent sandy tills overlaying fractured bedrock which near lake level is overlain by varved silts and clays. Kamic terraces composed of sandy to gravelly sediments are common in tributary valleys. Lake George bedrock is a mix of dominantly granitic gneisses, charnockitic gneisses, garnet-biotite-quartz-plagioclase gneisses, quartzites, metaanthrosites and metagabbros with smaller quantities of marble, calcsilicates, and amphibolites (Shuster et al. 1994).

The effects of repeated glacial encroachment and retreat in the Lake George watershed resulted in varying soil types (Table 1) and thus sub-watersheds exhibit different hydraulic conductivities. The greater deposition of sand in the south end resulted in East and West Brook exhibiting the highest infiltration rates while Shelving Rock, the only watershed monitored on the east side of the lake, is characterized by steeper slopes and shallow/exposed bedrock limiting infiltration rates. The remaining watersheds (English, Finkle, and Indian) have soils primarily composed of type B and C soils which exhibit intermediate infiltration rates.

The six major tributaries of this study were monitored for temperature and chemical composition between 2007 and 2010. Sampling occurred monthly between December and March with the sampling rate increasing to 2-week intervals between April and November. Temperature measurements were taken during baseflow conditions from March to December using a Raytek Mini Temp IR thermometer, which was routinely checked against National Institute of Standards and Technology (NIST) certified thermometers. Water temperature measurements were taken while either in the stream or on the stream bank with the angle primarily vertical. West, English, and Finkle Brook measurements were taken near the mouth of the streams while East, Shelving Rock, and Indian Brook measurements were taken within 0.5 km of the lake. Shelving Rock and Indian Brook were dominantly shaded by riparian vegetation between the sampling location and the lake; a wetland was located downstream of the East Brook sampling location. Raw temperature measurements are presented along with relative water temperature difference, calculated as the individual stream temperature minus the average temperature of all streams sampled on a specific date.

The Offshore Chemical Monitoring Program sampled mid-lake locations from May to November during 2007 and 2008. Prior to stratification, sampling occurred every 2 weeks with monthly sampling during summer and 2-week sampling reinstated during fall. The two locations utilized in the South Basin were Tea Island in the Caldwell Sub-basin and Dome Island in the Dome Island Sub-basin (Figure 1). Vertical temperature profiles were recorded at set intervals of 0, 1, 2, 3, 5, 10, 15, 20, 25, 30 m with additional measurements taken within depth ranges exhibiting large temperature changes. Measurements were recorded using a YSI temperature probe (various models), which was routinely checked against NIST thermometers. Samples were collected for analytical chemistry in the epilimnion (0–10 m; hose-integrated) and hypolimnion (1 m off the bottom; grab). Sodium, calcium, magnesium, and potassium were analyzed by atomic absorption spectrophotometry using a Perkin Elmer AAnalyst 5000 (Creed et al. 1991); chloride and sulfate were analyzed by ion chromatography using a Lachat 8000 QuikChem (Pfaff 1993). Alkalinity was measured by Titration Method 2320 B (Clesceri et al. 1989). Standard quality assurance and quality control protocols included blanks, duplicate samples, spikes, and external check standards every ten samples. Additional description of sampling methods for the stream and lake monitoring can be found in Swinton et al. (2015).

Water density was calculated for both the lake and streams using temperature and salinity (comprising sodium, chloride, calcium, magnesium, potassium, sulfate, and bicarbonate). Although both epilimnetic and hypolimnetic lake samples were taken and their calculated salinities were within 5% of each other, only the epilimnetic value was used in the calculation as it was more representative of the water column (0–10 m) than a grab sample 1 m off the bottom, which could be influenced by underflow. While bicarbonate was not measured directly during the 2007–2008 stream study, it is estimated as being equivalent to calcium on a weight basis. Based on preliminary West and Finkle Brook 2015 data from the Jefferson Project, calcium and alkalinity (reported as calcium carbonate) exhibited a strong correlation (r² > 0.93, N = 6) during baseflow conditions with calcium comprising 37–60% of the calcium carbonate. Therefore, bicarbonate was estimated as an equivalent mass of calcium. Alkalinity measurements were taken for the lake sampling locations and were included in the salinity calculations.

ρ_o, A, and B are based on temperature, C is a constant, and S is salinity (g/kg).

Stream intrusion depth was based on density calculations for each stream and the corresponding sub-basin, Caldwell or Dome Island. Isopleths were constructed using SigmaPlot and because density was interpolated between sampling dates and within individual profiles some static instabilities may be present. Isopleths contours are depicted using the oceanographic sigma or density anomaly measurement which is the water density difference from 1,000 kg/m³.

To verify the temperature trends observed in this stream study, previous stream studies that included West or East Brook and any of the other four streams included in the 2007–2010 project were examined for comparison. Fuhs (1972) sampled a total of 18 streams from July 1970 to July 1971; temperature data from West, English, Finkle, and Indian Brook are included here for comparison. The Nationwide Urban Runoff Program (NURP) focused on streams having different levels of watershed development at the south end of the lake from July 1980 to June 1982, and therefore only includes West and English Brook (Sutherland et al. 1983). Sutherland (unpublished data) sampled from August 2002 through November 2005 on West, East, English, Finkle, and Indian Brook.

A YSI 6950 vertical profiler equipped with EXO 2 sonde capable of recording temperature and chloride profiles at 1-m resolution every 90 minutes was deployed near the deepest location (53 m) in the Dome Island Sub-basin during the fall of 2014 as part of the Jefferson Project. The water temperature probe had a resolution of 0.001 °C with an accuracy of 0.01 °C between −5 and 35 °C. The chloride probe had a resolution of 0.01 mg/L with accuracy ±15% of reading or 5 mg/L between 0 and 1,000 mg/L. Air temperature was measured using the Vaisala Weather Transmitter WXT520 attached to the vertical profiler platform. The air temperature measurements had a resolution of 0.1 °C with an accuracy ranging from 0.2 at −20 °C to 0.4 at 40 °C. The Jefferson Project is a collaboration among Rensselaer Polytechnic Institute, IBM Research, and the FUND for Lake George with the goal of combining multiple high-resolution data from weather, stream, and in-lake sensors to create meteorological, hydrologic, hydrodynamic, and food web models to better understand the effects of anthropogenic development and climate change on the health and function of Lake George.

Statistical analyses were conducted in SPSS or SigmaPlot. Normality and equal variance were conducted on baseflow stream data to determine if parametric or non-parametric analyses were appropriate. Significance of main factors was determined using analysis of variance (ANOVA) with the Holm-Sidak method used for pairwise comparisons when data were normally distributed. Data that were non-normally distributed required the non-parametric counterparts: Kruskal–Wallis and Dunn's methods to determine significant differences between main factors and pairwise comparisons, respectively. Correlations were conducted using the Spearman rank order when data distribution was non-normal.

Seasonal components were based on solstice and equinox dates: spring (March 21–June 20), summer (June 21–September 22), and fall (September 23–December 20).

To reduce inter-annual variability and thus enhance statistical rigor, the relative water temperature differences were compared among streams. Comparing the relative stream temperatures strengthened the differences observed during summer and resulted in new significant comparisons during the spring and fall. Along with East and West Brook, significant summer water temperature differences now included Shelving Rock being significantly cooler than English, Finkle, and Indian Brook. During the spring, East, West, and Shelving Rock Brook were significantly (Dunn's method, p < 0.05) cooler than Finkle and/or Indian Brook. During fall, Shelving Rock Brook was significantly (Holm-Sidak, p < 0.05) cooler than West, English, and Finkle Brook.

In early May 2008, English Brook entered the lake near the surface and East Brook as underflow (Figure 5). Establishment of a strong thermocline was delayed allowing East, West, and English Brook to intrude at ∼15 m in mid-June compared to ∼10 m at the same time in 2007. East, West, and English Brook generally inserted between 10 and 15 m until August with a tendency for East Brook to intrude the deepest and English the shallowest. Through September, English maintained the 10 to 15 m intrusion depth with East and West entering between 15 and 25 m. By October, all were intruding as underflow.

Streams entering the Dome Island Sub-basin had a tendency to insert either above or within the metalimnion until late August or September during 2007 and 2008, respectively, with the exception of Shelving Rock in early 2007 (Figure 5). Finkle and Indian Brook inserted at the lake surface through May 2007 while Shelving Rock Brook tended to enter as underflow or interflow. Between June and early August, all streams inserted within the metalimnion, with intrusion below the metalimnion beginning in mid-August. By mid-September, stream inputs intruded as underflow. Similar patterns were observed during 2008.

Summer baseflow temperatures in East, West, and Shelving Rock Brook clearly were cooler than other major streams monitored, and based on historic West Brook studies, this is not a new phenomenon. Therefore, the discussion will focus on: (1) how soil composition may influence the different seasonal thermal regimes observed; (2) how deeper intrusion depths may incorporate organics and oxygen into the hypolimnion; and (3) the implications of hypolimnetic intrusion on the oxygen depletion regularly observed in the Caldwell Sub-basin.

Three distinct seasonal temperature patterns exist in the six major streams monitored around Lake George between 2007 and 2010 that can be explained in part by the soil composition within each watershed. The most common thermal regime is representative of English, Finkle, and Indian Brook. Median summer baseflow temperature among these streams was very consistent ranging from 17.3 to 17.5 °C (Figure 2). However the spring (average) and fall (median) temperatures differed by 1.7 °C and 1.1 °C, respectively. These streams are located along the west side of the lake with >83% of the soils comprising type B and C (Table 1) having an infiltration rate between 0.05 and 0.30 in/hr (Guo 2006). Since soil type and topography are similar, the influence of groundwater should be fairly consistent among the streams, as seen in the average summer temperatures. Stream temperatures in the spring and fall experience increased variation because the influence of groundwater and canopy cover diminish, enhancing the effect of climate and stream morphology.

The cooler summer temperatures in East and West Brook can be explained by the greater proportion of fast infiltrating soils; ∼75% of soils in these two watersheds comprise type A and B soils. These soils consist primarily of sand and gravel which result in infiltration rates between 0.15 and 0.45 in/hr (Guo 2006); the faster infiltration rate means groundwater seepage into the streams can occur more readily. The assumption that a greater proportion of discharge from these streams originates from groundwater is supported by the cooler water temperatures in warm months and warmer water temperatures in cold months. Since groundwater in the region remains ∼8.3 °C throughout the year, it acts as a temperature buffer during summer and winter. The interaction between stream water and groundwater is well documented (Castro & Hornberger 1991; Stanford & Ward 1993; Evans et al. 1998; Malcolm et al. 2002) along with its ability to buffer stream temperatures (Holmes 2000; Malcolm et al. 2002; Johnson 2004). One of the most compelling studies by Shepherd et al. (1986) combined three studies along the Pacific Northwest coast of the United States varying in watershed size, discharge rate, and temperature during different decades using different methodologies and having different goals: all documented that intra-gravel temperature was warmer in the winter and cooler in the summer than the stream temperature with the transitions occurring around March and October.

Shelving Rock Brook, the only stream monitored on the east side of the lake, is characterized as undeveloped forest with steeper slopes and a greater proportion of shallow/exposed bedrock relative to the other watersheds monitored. Shelving Rock Brook temperatures generally were cooler than the other streams with the exception of East and West Brook during summer. Shelving Rock Brook temperatures did not exhibit a seasonal component when analyzing relative temperature differences, indicating groundwater inputs were not the primary influencing factor. While the heavily forested watershed may aid in maintaining cooler water temperatures, it is likely not a principal factor because the Indian Brook watershed also is heavily forested but exhibited a different thermal regime. The most probable explanation is the greater proportion of shallow/exposed bedrock. Shallow streams with bedrock bottoms can transfer up to 25% of the energy absorbed by the streams to the bedrock resulting in a dampening of diurnal temperature (Brown 1969). Additionally, the stream corridor is predominantly shaded providing insulation by limiting energy absorption from solar radiation. Small stream temperatures are difficult to predict because they respond more rapidly to energy inputs and watershed characteristics than do larger streams (Smith 1972), making it necessary to have continuous temperature measurements and energy flux calculations among the streams and their substrates to fully understand the different thermal regimes observed throughout the Lake George Basin.

As stream discharge enters a lake it will intrude according to its density relative to the water column. If the stream discharge is less dense than the lake, the discharge enters as overflow and becomes incorporated into the surface mixed layer (SML). This was the case for Finkle and Indian Brooks through late spring of 2007. If the stream discharge is denser than the lake, as is often the case for East, West, and Shelving, the flow will propagate down the slope of the lake. As the flow continues down the slope of the lake, a head begins to form at the front of the flow; it is this region of the flow that mixes with the lake water (Simpson 1982). The extent of mixing and entrainment of the head depends on the velocity, density difference, and the degree of slope. As any of these parameters increase, mixing between the head of the flow and the lake water also increases (Simpson 1982). The flow will continue down the slope until it reaches a depth of neutral buoyancy, at which time it begins to propagate horizontally (Alavian et al. 1992). Prior to stratification, the water column is well mixed and while East and West Brook may intrude at varying depths, all inputs will likely become mixed throughout the water column.

Once the thermocline develops, intrusion depth will dictate the region of the water column where the discharge will incorporate. If the discharge intrudes above the metalimnion, the flow will become incorporated into the SML (Cortés et al. 2014). However, if the intrusion is within the metalimnion, it likely will not become incorporated into the SML quickly, due to the rate of vertical mixing within a thermocline being on the order of heat diffusion (Quay et al. 1980; Fee et al. 1994). Therefore, stream discharge intruding into the metalimnion will dominantly remain in the metalimnion for an extended period of time, possibly creating a nutrient-rich layer for primary producers to utilize. If the discharge is denser than the metalimnion, the flow continues to propagate down the slope into the hypolimnion (Cortés et al. 2014). East and West Brook intrusion depths during the summer and fall imply the inputs would predominantly be maintained in the metalimnion or hypolimnion with little incorporating into the SML until fall turnover.

High-resolution profiler data confirmed that underflow does occur during fall at the deepest depths of the Dome Island Sub-basin. Initial data identified a colder, less saline pulse of water entered the sub-basin during a storm event in November 2014 (Figure 7). Streams entering the lake in this area have watersheds that primarily are forested and are representative of Shelving Rock Brook, which is located ∼1.5 km away from the profiler. Shelving Rock Brook chloride concentration was ∼1 mg/L (Swinton et al. 2015) with November baseflow temperatures measured between 2007 and 2010 ranging from 0.6 to 6.6 °C. The colder and less saline inputs entered the lake as underflow with entrainment affecting water below 40 m; little temperature change was detected in the top 40 m of the water column. The high-resolution data illustrate the power to identify complex hydrodynamic mixing during storm events. An additional profiler was deployed in the Caldwell Sub-basin, near East and West Brook during the 2015 season, which should allow us to test the hypothesis that cooler stream temperatures insert into the hypolimnion as implied by the calculated intrusion depth. The high-resolution data can additionally identify if intrusion depths vary throughout the day and how temperature changes during summer storm events influence intrusion depth and mixing patterns, assuming that the inputs during baseflow conditions can be detectable at these offshore locations. The inputs may simply be too small to detect and, if that is the case, installation of static CTD sensor strings near shore could benefit future studies.

The depth of stream intrusion is of great importance when studying or managing a lake because the nutrients, sediments, gases, and pollutants incorporated in the discharge will be available at specific levels of the lake during periods of stratification. The primary concern with deeper intrusion depths in the Caldwell Sub-basin of Lake George deals with seasonal hypolimnetic oxygen depletion. In the Caldwell Sub-basin, the hypolimnion regularly experiences oxygen depletion below 4 mg/L during late summer/early fall (Boylen et al. 2014). This is the only sub-basin of the lake where oxygen depletion this severe has been documented and may be impacted by deep intrusion of nutrient-rich oxygenated stream discharge. East and West Brook are two sub-watersheds at the south end of Lake George with high residential and commercial development associated with tourism. West Brook has been known since the late 1960s–early 1970s to contribute elevated levels of nitrogen and chloride (which also affects water density) to the lake that originate from the local waste water treatment plant (Aulenbach & Tofflemire 1975) and a previously uncovered road salt storage facility. Since at least the early 1980s, East Brook and an adjacent small sub-watershed have contributed high levels of phosphorus to Lake George through stormwater runoff (Sutherland et al. 1983). When these nutrient-rich oxygenated inputs are incorporated into the hypolimnion, they can have both positive and negative effects on the dissolved oxygen levels. The addition of organic material could negatively affect the oxygen levels by elevating the microbial activity and thus promote hypolimnetic oxygen depletion. On the other hand, the input of oxygen-rich stream discharge could attenuate the seasonal hypolimnetic oxygen depletion. An in-depth study on how stream discharge impacts hypolimnetic oxygen concentration is required to determine the extent and importance of supplying both organics and oxygen to the hypolimnion during stratification.

Soil type variability within the Lake George watershed resulted in distinct stream temperature regimes that influenced intrusion depth into the lake. Streams receiving a large proportion of groundwater or more shallow/exposed bedrock exhibited cooler summer temperatures dictating deeper intrusion depths that could isolate stream inputs to the hypolimnion during periods of stratification. High-resolution profiler data confirmed underflow intrusion during the fall of 2014 and merit additional data collection to determine if the stream discharge is able to penetrate the thermocline to incorporate into the hypolimnion. If so, this deeper intrusion could have both negative and positive impacts on the hypolimnetic oxygen depletion that occurs in that sub-basin of the lake. The addition of high-resolution profiler data throughout the lake as part of the Jefferson Project will answer questions similar to these and progress our understanding of the hydrodynamics in Lake George as well as the field of hydrodynamics.

This research was a combination of several studies with separate funding sources. The 2007–2010 stream study was funded by The Lake George Watershed Coalition. The DFWI Offshore Chemical Monitoring Program has been jointly funded by Rensselaer Polytechnic Institute and the FUND for Lake George. The high-resolution profiler data from The Jefferson Project, a collaboration between Rensselaer Polytechnic Institute, IBM Research, and the FUND for Lake George, was made possible through financial contributions of its participants. The DFWI is grateful for the steadfast support of the David M. & Margaret A. Darrin and United Parcel Service endowments.