1. Introduction
In terms of surface area, Great Slave Lake (GSL) and Great Bear Lake (GBL) are, respectively, the 10th and 8th largest freshwater lakes in the world and the two largest at such high latitudes (Fig. 1). GSL and GBL have in common that they are large, deep, and cold (Fig. 2, Table 1). Both have very short daylight periods in winter and equally long ones in summer. They are, however, dissimilar hydrologically. The open-water period for GSL is substantially longer than for GBL and the water surface temperatures during that period are substantially higher. The drainage area of the GSL subbasin is more than twice as large as that of the GBL basin and its outflow discharge into the Mackenzie River is more than 8 times that from GBL. A comparison of the volume flow records of the Mackenzie River just downstream from GSL, and the Slave River before it enters GSL, indicates that 82% of the discharge from GSL is derived from inflow from the south via the Slave River and 18% from the GSL subbasin. The importance of this southern catchment to the hydrology of GSL is clearly outlined in Gibson et al. (2006). All of the discharge from GBL derives from its local watershed. It will be apparent from the results of this presentation that the position of GSL as part of the Mackenzie River flowthrough system, compared with the more hydrologically isolated position of GBL (Fig. 1), plays a substantial role in the markedly different thermodynamic–hydrodynamic characteristics and surface climates of the two lakes.
This study is the first to compare the two lakes. A sustained research effort has been devoted to GSL over the last decade as part of the Mackenzie Global Energy and Water Cycle Experiment (GEWEX) Study (MAGS; e.g., Blanken et al. 2003; Rouse et al. 2003a, b; Schertzer et al. 2003), but research on GBL has been minimal. For GBL, substantive work was done on lake temperatures by Johnson (1975, 1994), but subsequent efforts were not undertaken until the present century and it is these that are incorporated into this report. Our basic hypothesis is that because of their similar high-latitude positions, substantial sizes, depths, and volumes, GSL and GBL will exhibit similar surface and near-surface climates and are typical of large lakes in the high latitudes. This is an important issue because large lakes are frequent in circumpolar high and middle latitudes, especially in the Canadian Shield region of North America (Oswald and Rouse 2004). Hence, knowledge of their hydrodynamic and thermodynamic behavior is important to understanding regional climates and to regional climate modeling (Rouse et al. 2005).
Most of the measurement techniques have been described elsewhere and here are only outlined briefly. We document the representativeness of our measurements since, in such a data-sparse area, this is critical to addressing our basic hypothesis. A comparison between the thermodynamic and hydrodynamic characteristics and surface climates of GSL and GBL is emphasized. Results are presented in subsections that introduce the regional climatic regimes, the thermal regimes above and within the lakes, the seasonality of the heat and water cycles, temperature and water vapor gradients at the lake–atmosphere interface, and the surface energy balances. Our discussion examines the potential errors in the results, addresses the relevance of the results, and assesses the validity of our hypothesis.
2. Methods
a. Research sites
Two small midlake islands, one each in Great Slave Lake (GSL) and Great Bear Lake (GBL), served as research stations from which to examine the interactions of the lakes with their atmospheric environments (Fig. 2).
On GSL, Inner Whaleback Island (61.9°N, 114.7°W) has recently been established as a Meteorological Service of Canada network site. This small rocky island is located in the main basin 12 km from shore, 70 km southeast of Yellowknife, and 150 km northeast of Hay River (Fig. 2). The surrounding water depth is close to the mean lake depth of 50 m and the site is well located to represent general midlake processes (Blanken et al. 2000, 2003). This location was employed for 6 yr as a research base to study midlake processes. The island’s height above the mean water surface, width, and length are approximately 10, 100, and 180 m, respectively, with the long axis oriented east–west.
On GBL, Lionel Island (65.5°N, 122.0°W; Fig. 2) was employed for 2 yr as a research base. It is situated about 200 km east of Norman Wells (Fig. 1) and 60 km east of Deline (Figs. 1 and 2). Within the lake, it is approximately positioned at the 50-m-depth contour, which is representative of the moderately deep parts of GBL (Fig. 2). The island’s height above the mean water surface, width, and length are approximately 2.5, 50, and 170 m, respectively, with the long axis oriented northeast–southwest.
b. Measurement periods
Most of the measurements at GSL have been averaged over 5 yr between 1997 and 2003, and those at GBL over 2 yr, 2004–05 (Table 2). Solar radiation was measured at Inner Whaleback Island and Lionel Island for 4 and 2 yr, respectively. Measurements from the meteorological stations of Hay River and Yellowknife were averaged to give regional air temperatures representative of GSL. Data from Norman Wells were used to depict the regional temperatures for GBL. This provides 63 yr of comparative regional temperature data for the two research locations. Ice thickness measurements were taken in the winters of 2003/04 and 2004/05 for each lake.
c. Regional air temperature
Regional air temperatures offer one of the important controls when comparing sites. Temperatures at Norman Wells (65.3°N, 126.9°W), which lies 140 km to the west of GBL (Fig. 1), are taken to accurately represent temperatures in the region of GBL. Hay River (HR; 61.3°N, 115.7°W) and Yellowknife (YK; 62.5°N, 114.4°W), however, are nearshore sites on the south and north shores of GSL, respectively (Fig. 1). The bias and temperature adjustments due to these nearshore locations are determined as follows. It is assumed that when winds blow from the east and west, which, averaged over the year, is about half of the time, there is no significant effect from GSL on the measured temperatures at each station. Under these wind directions, the measured temperature accurately represents the regional temperature. When the wind blows from the north and south, which totals about half of the time (Table 3), the measurements at HR and YK are differentially affected. This influence varies with wind direction and the condition of the lake (i.e., frozen, thawing, open, or refreezing).
Details of the temperature bias correction (TaTBC) are outlined in Table 3. Positive values for TaTBC denote that GSL is exerting a warming effect and negative values that it is exerting a cooling effect. The annual effect of the wind blowing across GSL is to raise temperatures at the nearshore stations of HR and YK by 0.4°C (Table 3). On a monthly basis this effect varies between warming of up to 1.0°C and cooling of up to 0.2°C. Regional air temperatures have been adjusted for these biases (Table 3).
d. Air temperature at the island research sites
Air temperature and relative humidity measurements at Inner Whaleback Island and Lionel Island employed shielded sensors (Vaisala HMP-45C, Campbell Scientific, Logan, Utah) at a height of 1.5 m. The output was logged at 5-s intervals and integrated over 10-min intervals. These two island sites are used to represent overlake air temperatures. They are also used in conjunction with surface water temperature measurements at the nearby lake thermistor strings to establish daily averaged boundary layer air temperature and vapor pressure gradients.
e. Midlake water temperatures, heat storage, and solar radiation
Water temperature moorings, deployed near each of the research islands, were equipped with self-recording Stowaway Tidbit temperature loggers. Sensor depths were standardized at 0, 5, 10, 15, 20, 25, 30, and 40 m to allow for comparability of thermal profiles. Observations were recorded at 15-min intervals. The surface temperature was measured with a floating, shielded thermistor attached to the mooring buoy. On average, the thermistor string measurements covered a period of 178 days [day of the year (DOY) 167–344] for GSL and 33 days (DOY 181–213) for GBL. Because of varying ice conditions and other factors, the times of employment and deployment varied from year to year. National Oceanographic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) satellite surface water temperatures were calculated for 1999 using the methodology describe in Bussières and Schertzer (2003). The average temperatures calculated for the two lakes represent a large portion of their water surfaces (18 500 km2 for GSL and 25 300 km2 for GBL). A 5 km × 5 km grid centered on each of the research islands is used to determine local water surface temperatures for comparison with the lake-wide average. The AVHRR analysis is used to establish the representativeness of the in situ lake measurements to the overall lake surface temperatures. This is very useful in interpreting time-limited in situ measurements for GBL especially.
Incoming solar radiation was logged continuously at the two research islands using Eppley pyranometers (model 8–48).
f. Breakup, freeze-up, and lake ice
Passive microwave imagery can discriminate between ice cover and open water, and has proved well suited to large lakes (Walker et al. 2000). It is applicable at any time of the day, in all weather conditions, and provides a resolution of 12.5 km. Breakup and freeze-up are defined from the 37-GHz brightness temperature time series from the Nimbus-7 Scanning Multichannel Microwave Radiometer (SMMR; 1979–87 breakup) and Defense Meteorological Satellite Program (DMSP) Special Sensor Microwave Imager (SSM/I; 1987 freeze-up through 2005) passive microwave satellite sensors. During breakup, there is a very distinctive almost linear response during a period averaging about 2 weeks from the high winter snow–ice brightness temperatures in the range of 225 K, to summer ice-free ones in the range of 150 K. A similarly distinct transition occurs, in reverse, during freeze-up. The date of breakup is defined by the onset of the low summer brightness temperature time trace and date of freeze-up by the onset of the high winter snow–ice brightness temperature time trace.
Lake ice thicknesses were measured with ice augers. In the winters of 2004/05 and 2005/06, the measurements for GSL were made at Back Bay, which is adjacent to Yellowknife (Fig. 2). For GBL they were made at four different places in the southwest part of the Keith Arm. One of these sites was adjacent to Lionel Island.
The one-dimensional thermodynamic Canadian Lake Ice Model (Duguay et al. 2002, 2006; Ménard et al. 2002) is used to simulate ice thicknesses in the Mackenzie basin, including GSL and GBL. The model employs long-term average meteorological observations obtained at weather stations such as Hay River, Yellowknife, and Norman Wells.
g. Latent and sensible heat fluxes
Two different eddy covariance systems were employed on GSL. For the first 2 yr, an Mk2 Hydra (Shuttleworth et al. 1988), which employs a one-dimensional sonic anemometer was used and for the final 3 yr, a system with a three-dimensional sonic anemometer (model CSAT-3, Campbell Scientific) and an open-path gas analyzer (model LI-7500, Li-Cor Inc., Lincoln, Nebraska) was employed (Blanken et al. 2003). This latter system was used during the two measurement years on GBL. On GSL, Blanken et al. (2000) found that 80% of the turbulent flux measurements were sampled over an upwind distance extending at least 5 km over water (flux footprint). On GBL the fetch was unlimited in all directions except along the long axis of Lionel Island; a relatively rare occurrence for which the data were deleted from the analysis. The lower sensor heights (∼7.5 m above the water surface) and differences in atmospheric turbulent fluxes and conditions compared to GSL resulted in a smaller flux footprint. For example, the upwind distances to which the turbulent fluxes were most sensitive occurred at 84, 124, and 252 m during typical unstable, stable, and neutral atmospheric stability conditions, respectively. Eighty percent of the turbulent fluxes originated within 0.7, 1.1, and 2.3 km of the site (unstable, neutral, and stable atmospheric stability). For GBL, data in the fall and early winter periods were limited due to site access difficulties and a number of instrument problems. To obtain representative seasonal totals, time-dependent polynomial curves were fitted to the measured data.
3. Results
a. Regional air temperature and solar radiation
A number of observations emerge from analysis of the regional air temperature (Fig. 3). Annually, the long-term (1943–2005) regional air temperatures for GSL and GBL were −3.8° and −5.5°C, respectively, and for the open-water period (May–December) they were 2.7° and 1.3°C, respectively. For the years of measurement detailed in this report (Table 2), the regional temperatures annually and during the open-water period were within 0.4°C of the long-term average for both lakes. Thus, regional temperatures in the measurement years were fully representative of the long-term averages.
The incoming solar radiation (Fig. 4) represents different years at each lake (Table 2) so that magnitudes are not directly comparable. During the open-water period, incoming solar radiation for GSL exceeded that for GBL by 10%. GSL shows dampened fluctuations over time, probably due to a 4-yr averaging period as opposed to 2 yr for GBL. It is significant that the average time of ice breakup on GSL occurs around the same time as the summer solstice, but ice breakup on GBL lags, occurring about 3 weeks after the summer solstice (Fig. 4). The importance of this to the magnitudes of absorbed solar radiation is addressed later in this paper.
b. Overlake thermal regimes
It is well established that large lakes exert a strong influence on their overlake air temperatures. We are in a position to compare these effects for the two large high-latitude lakes. Both GSL and GBL have overlake temperature regimes that differ substantially from their regional temperatures (Figs. 5a and 5b). The pattern for GSL is similar to that described by Rouse et al. (2003a). Both lakes are cooler than their surroundings during spring–summer and warmer during fall–early winter. The transition from the lakes being cooler to being warmer than their surroundings comes in mid-September for GSL and in late September for GBL. GBL shows larger magnitudes of difference in regional to overlake air temperatures than is measured for GSL in both the “lake colder” and “lake warmer” phases. GSL has higher overlake air temperatures than GBL from late May to early October, but from then onward to the year’s end there are no significant differences between the two lakes in average overlake temperatures (Fig. 5c).
c. Midlake water temperatures and heat storage
The temporal pattern of midlake water temperatures at the surface (Tw0) and at the 40-m depth (Tw40) indicate that GSL is much warmer than GBL at almost all times (Figs. 6 and 7). The averages at GSL for Tw0 and Tw40 are 7.5° and 4.1°C, and for GBL they are 4.8° and 3.5°C, respectively. Employing a degree-day calculation (one degree day is assigned daily for each 1°C > 0°C) gives degree-day totals at the surface and at the 40-m depth of 1449 and 844 at GSL and 633 and 468 at GBL. These values are a good indicator of the relative heat content of the two lakes at these two depths during the open-water season.
The major differences in midlake surface water temperatures for the two lakes are supported by evidence from the AVHRR analysis for 1999, which indicates also that the midlake temperature differences are indicative of the overall lake surface temperature differences (Fig. 6).
d. Breakup, freeze-up, open water, and ice
The time series analysis (Fig. 8) derived from passive microwave sensors indicates that, for the 20-yr period from 1979 to 2005, the average dates for ice breakup and freeze-up for GSL were 16 June and 9 December, respectively. The comparable dates for GBL were 13 July and 29 November (Fig. 8; Table 4). This gives average open-water periods of 175 and 139 days for GSL and GBL, respectively. There is a large variability about the averages (Fig. 8). The difference between years of the maximum and minimum length of open water for both GSL and GBL is 56 days (Table 4a). For the years with the maximum length of open water, each lake thawed about 3 weeks earlier and froze from 2–3 weeks later than average (Table 4b). For the years of minimum length of open water, each lake thawed about a week later and froze from 1 to 2+ weeks earlier than average. This emphasizes the role of (early) late thawing in (enhancing) limiting the absorption of solar radiation by open water, which, in turn, is primarily responsible for maintaining or limiting the length of open water into late fall and early winter (Rouse et al. 2003a, 2005). It also indicates a similar climate response during extreme years with respect to ice between the two lakes, even though they are at significantly different latitudes. This is particularly notable for the El Niño year of 1998 where early thaw and late freeze-back at both lakes gave the longest open-water periods in the entire records for each.
For comparable data periods in the winters of 2004 and 2005, maximum ice thicknesses on GBL averaged about 0.20 m more than on GSL (Table 5). Differences in ice thickness between the lakes varied for the 2 yr. In 2004, ice on GBL was 0.50 m thicker but in 2005 ice thicknesses were the same. GBL maintained its ice cover for 27 and 34 days longer than GSL in 2004 and 2005 (Fig. 8). Maximum ice thickness differences in the two lakes explain neither the final thaw dates, nor differences in the dates when the maximum ice thickness were reached. Results from the Canadian Lake Ice Model using average meteorological data accurately matched the 2-yr average ice thickness results for GSL and slightly underpredicted (by 0.1 m) for GBL (Table 5). The modeled maximum ice thickness on GBL occurred in mid-May, 40 days later than on GSL, a time difference that corresponds to the differences in the dates of the final ice melt (Fig. 8).
e. Midlake heat storage and gradients at the lake–atmosphere interface
As shown in Fig. 9, the midlake heat storage and heat exchange in GSL (2150 MJ m−2) was substantially greater than that in GBL (1046 MJ m−2). Maximum heat storage was achieved later in GSL than in GBL (Fig. 9). As a result, the heat loss from GSL was sustained much later into the winter period. The heating and cooling cycles of GSL persist for 155 and 79 days, respectively; whereas for GBL, the comparable cycles are 53 and 51 days (Fig. 9). Our data indicate that GSL’s warming cycle persists 3 times longer than that of GBL. Considering all of the depths in the water column, GSL heats twice as slowly as it cools, whereas GBL heats and cools at about the same rate. However, important differences in the rate of change in temperature are evident at different lake depths.
For much of the heating period in both lakes, the surface temperatures are lower than the overlying air and the vertical temperature and humidity gradients approach 1°C m−1 and 20 Pa m−1, respectively (Fig. 10). Under these inversion conditions, low-lying cloud or fog prevails. Conversely, during the cooling period, lapse conditions prevail. They too can be large in the late fall and early winter period, sometimes exceeding −1°C m−1 and −30 Pa m−1. These large gradients promote vigorous downward-directed atmospheric convective fluxes early in the warming phase and upward-directed fluxes late in the cooling phase.
f. Latent and sensible heat fluxes
A number of features of the seasonal trends in the latent (Qe) and sensible (Qh) heat fluxes would be anticipated from the seasonal heat storage patterns. For GSL, there is a notable increase in the magnitudes of heat loss after mid-September (DOY 260) that increases into late fall and early winter (Fig. 11). For GBL, there are no measured flux data for the late fall–early winter period, but indirect evidence indicates that the magnitude of the heat loss is small since the midlake representative water column is at or near its minimum heat content (Fig. 9). The magnitudes of Qe and Qh are substantively larger for GSL than for GBL (Fig. 12). The heat flux, Qh, is an important component of the seasonal convective heat exchanges with bulk seasonal Bowen ratios (= Qh/Qe) of 0.61 and 0.82 for GSL and GBL, respectively.
4. Discussion
a. Error assessment
When comparing results from two very large lakes separated geographically by 200 km, over the disparate timeframes shown in Table 2, with limited measurements, there is considerable potential for errors. The probable magnitudes of these errors are treated both qualitatively and to some extent quantitatively (Table 6), and their probable impacts on the results are assessed.
Regional air temperatures are based on standard measurements at airport weather stations that have been in operation for a long period. The adjustment for bias due to the cooling effect of GSL resulting from the nearshore position of the two stations does not exceed 1°C in any given month (Table 3) and over the full open-water period constitutes a modest 0.3°C month−1. The bias adjustment is designed to give more realistic regional temperature comparisons (Fig. 3).
Differences in overlake temperatures and regional air temperatures for the two lakes (Fig. 5) rely on the representativeness of the measurements at the island site in each lake as well as on the regional temperatures discussed above. Both island sites employ the same standard measurement techniques as at the airport sites.
Water temperature measurements apply only to the water column represented by the thermistor strings and cannot claim lake-wide representativeness. They also represent different measurement years for the two lakes so that direct temporal comparison is not achieved. Additionally, the measurements for GBL are seasonally limited (Table 2). All of these factors restrict the interpretation of the comparative results. On the positive side, there is good comparison between contemporary AVHRR lake surface temperature calculations at the two lakes with the observed surface temperatures made during the measurement years at each site (Fig. 6). This indicates that substantial differences in the magnitudes of water temperatures (Fig. 7) are a consistent feature of the two lakes and that this is consistent with the substantial differences in the dates of the final breakup and final freeze-back (Fig. 8).
Heat storage calculations apply to the water columns noted above. Their accuracy is determined by the accuracy and length of the measurement records. Because of the sustained record for GSL, the polynomial projections are considered to be quite reliable. During the measurement years GSL received more incident solar radiation than GBL and displayed an 8 day longer open-water period than normal when compared to GBL. We estimate that the combined factors account for 13% of the calculated difference in maximum heat storage between GSL and GBL. We place the overall uncertainty in heat storage calculations at ±18%. For GBL the measurements are limited primarily to midseason data. It is estimated that the polynomial projections for the early and latter part of the open-water period could create a potential error of ±25% in the calculation of maximum heat storage. The combined potential errors result in a ±31% RMS error when comparing the maximum heat storage in the two lakes (Table 6). Errors in the eddy covariance calculations of the latent and sensible heat fluxes are larger for GBL than for GSL because of the longer period that the polynomial projections of these fluxes are applied to encompass the open-water season (Fig. 11). Table 6 indicates the following for the open-water period. The regional temperatures are very similar, with the differences being of similar magnitude to the potential errors. There is a significant difference in overlake air temperatures: those of GSL being higher than those of GBL. Potential errors in the calculation of lake water temperatures are relatively large, but the measured differences indicate that GSL is substantially warmer than GBL, which is also indicated in the AVHRR comparisons. Midlake heat storage in GSL is significantly greater than in GBL. The convective heat fluxes, Qe and Qh, from GSL are much greater than from GBL.
b. Significance of results
Both lakes exhibit similar regional temperature regimes. Because of a larger incident solar radiation and a longer low-albedo, ice-free period, the absorbed solar radiation (K*), integrated over the thawing and open-water period, is 486 MJ m−2 larger for GSL (1582 MJ m−2) than for GBL (1096 MJ m−2). Here, K* (GSL) can account for about 80% of GSL’s midlake maximum heat storage and K* (GBL) can account for almost all of GBL’s midlake maximum heat storage.
The results in Figs. 3, 5 and 8 are brought together in Table 7. The period from mid-May to the end of December includes lake heating and lake cooling periods. When the lakes are warming, they exert a substantial cooling effect on the overlying atmosphere and, conversely, when cooling, they exert a warming effect. These influences apply to both the frozen and open-water phases. During lake heating the relative cooling of the overlying atmosphere is substantially greater over GBL than over GSL. During lake cooling the magnitudes for the two lakes are similar during the open-water phase but diverge after freeze-up. It is readily apparent why GBL exerts larger local cooling during the early season heating phase. The regional temperatures for GSL and GBL increase at about the same rate, but GBL remains frozen or very cold. Strong temperature and humidity inversions promote both overlake cooling and condensation, the latter giving rise to a persistent low cloud.
The later ice breakup on GBL results in a large heat energy deficit relative to GSL. This becomes most important to the comparative hydrodynamics and thermodynamics of the two lakes. On both lakes, early seasonal thaw is accompanied by late seasonal freeze and late thaw by early freeze (Fig. 8). The early thaw–late freeze scenario derives from more absorbed solar radiation in spring promoting greater heat storage during the heating phase. In turn, this necessitates a greater heat release to the atmosphere in fall–early winter, which prolongs the cooling period and results in a later freeze. The system works in reverse for the late thaw–early freeze scenario.
The temperature and humidity gradients at the lake–atmospheric interface (Fig. 10) are closely tuned to the thermodynamics of the lake. The greater midlake heat storage in GSL (Fig. 9) results in higher surface temperatures and saturated vapor pressures, which persist for longer periods. These promote large vertical temperature and vapor pressure gradients that become especially prominent in fall and early winter. Vigorous turbulent mixing fueled by strong winds combines with these steep gradients to create large latent and sensible heat fluxes and large evaporation.
Differences in the length of the open-water period for GBL and GSL, as well as differences in the magnitudes of heat storage and the convective heat fluxes, cannot be accounted for by differences in incident solar radiation and regional air temperatures alone. Most of these differences are triggered by the different hydrologies of the two lakes.
The waters flowing into GBL originate at similar latitudes under the same regional climate as the lake. Inflow and outflow magnitudes are small relative to lake volume. Thus, the surface water exchange exerts a small impact on GBL, and its thermodynamics are largely determined by the interaction of its mixing layer with solar radiation and with the atmosphere above. It is a locally driven system.
In contrast, much of the water flowing into GSL has its origin in the main range of the Rocky Mountains from as far south as 53°N. The streams and rivers flow east then north through the Great Plains in major systems such as the Peace, Athabaska, and Slave Rivers. The southern input is also fed from Lake Athabaska before it enters GSL via the Slave River. As noted above, 82% of the total advective water exchange in the lake originates in this southern catchment, which becomes the major contributor to an outflow more than 8 times that of GBL. In the region of the southern catchment, prespring thaw and spring thaw experience 60% and 18% more degree days, respectively, compared with the GSL local catchment area. River breakup occurs 18 days earlier than lake breakup at GSL and freeze-up occurs 9 days later. Both before and after the thaw on GSL, the input from the southern catchment adds substantial heat to the lake. We know, from satellite images and from model results (Leon et al. 2005; Schertzer et al. 2008), that most of the water entering GSL from the Slave River flows in a counterclockwise gyre through the central basin including Inner Whaleback Island before exiting into the Mackenzie River at the lake’s western end. Most of this low-density warm current remains near the surface and promotes earlier ice thawing and higher surface temperatures than can occur in a more isolated lake such as GBL.
A factor that undoubtedly plays a role in the difference in the surface energy balances is the effect of storm track incidence accompanying the Arctic annular mode (e.g., Thompson and Wallace 2001; Chang and Fu 2002). This could directly affect evaporation from the two lakes due to their 5° latitudinal difference. For example, the passage of storm systems over lake waters results in a strong mixing of surface waters with deeper layers (Schertzer et al. 2003), and passages of cold dry air masses over large lakes, such as GSL, result in greatly enhanced evaporation (Blanken et al. 2000; Blanken et al. 2008). Additionally, the extensive southern watershed of GSL is subjected to different storm track frequencies than the immediate watershed of the lake. Such an extensive north-flowing river system as the Mackenzie is subjected to latitudinally variable meteorological regimes that will differentially influence the hydrology and energy balance of the large lakes.
Our hypothesis that because of their high-latitude position, similar regional climate, and substantial sizes, depths, and volumes, Great Slave Lake and Great Bear Lake will exhibit similar surface and near-surface climates and be typical of large lakes in the high latitudes is clearly invalid. This is primarily due to their very different hydrological regimes. The two lakes may represent extremes. GSL has very large through-flow and incorporates water originating from up to 8° latitude farther south, whereas Great Bear Lake is isolated in a relatively small local watershed.
It is well established that GSL responds rapidly and vigorously to climate variability (Blanken et al. 2003; Rouse et al. 2003a). The most evocative evidence from GSL is for the El Niño year of 1998 when exceptionally warm conditions led to an early melt, a late freeze-up (Fig. 8), and a large storage of heat during the summer. High lake temperatures stimulated large fluxes of latent and sensible heat that persisted almost to the year’s end. Since GBL underwent a similar lengthening of the open-water period due to early thaw and late freeze-up (Fig. 8), it is logical that the thermally induced early melting of the GBL lake ice would have enhanced the heat storage from absorbed solar radiation and increased the overall lake energy level. That, in turn, would have stimulated larger convective heat fluxes associated with later freezing of the lake. A similar feedback is identified by Austin and Colman (2007), indicating that Lake Superior’s surface waters are warming faster than regional air temperatures because of decreases in the length of the ice-covered season. On a larger scale, a similar feedback relationship has been postulated for early ice breakup in the polar sea (Serreze and Francis 2006).
5. Conclusions
In spite of the facts that both Great Slave Lake (GSL) and Great Bear Lake (GBL) are large, cold, and deep and have similar regional climates, the hypothesis that GSL and GBL will exhibit similar surface and near-surface climates that are typical of large lakes in the high latitudes is invalid. This arises because of their dissimilar hydrological regimes. All of GBL’s inflow derives locally from its relatively small watershed. The drainage area of the GSL subbasin is more than twice as large as that of the GBL basin and its outflow into the Mackenzie River is more than 8 times as large. Of this discharge, about 82% is derived from inflow from the southern basin and 18% from its local subbasin. GSL is very strongly influenced by the influx of relatively warm waters from the Peace–Athabaska–Slave River system to the south, an influence that is especially noteworthy in the spring, when GSL experiences breakup close to a month in advance of GBL. This breakup occurs close to the summer solstice when large inputs of solar radiation are absorbed by the low-albedo lake waters during the long summer days. The results are pronounced. GSL achieves higher lake temperatures, stores substantially more heat energy, and exhibits larger convective heat fluxes compared with GBL.
Several observations apply to both lakes. During the period of lake warming, each lake exerts a substantial local cooling effect on its boundary layer atmosphere and during cooling a substantial warming effect. This influence is greater for GBL. Temperature and humidity inversions are frequent early in the warming season and very strong lapse gradients occur late in the season. Thus, the evaporation is small or negative in the period after breakup and very large late in the period prior to freeze-up. For both lakes, early thaw is accompanied by late freeze-up and late thaw by early freeze-up. Early thaw promotes more absorbed solar radiation during the high-sun season. The stored energy from this source keeps the lakes open later into the winter. This relation between the time of lake ice breakup and the magnitude of the absorbed solar radiation makes the lakes very responsive to climatic variability.
Acknowledgments
Financial support for this project was provided by network research grants from the Meteorological Service of Canada (MSC) and the Natural Science and Engineering Research Council of Canada (NSERC) provided to the Mackenzie GEWEX Study, individual NSERC research grants, a research grant from NASA, and Student Northern Training Grants from the Canada Department of Indian and Northern Affairs. Logistical and equipment support during measurement campaigns has been provided by the Great Slave Lake branch of the Canadian Coast Guard; Environment Canada, Yellowknife, Northwest Territories; Institute of Hydrology, Wallingford, United Kingdom; Meteorological Service of Canada, Edmonton, Alberta, and Downsview, Ontario; National Water Research Institute, Burlington, Ontario; Polar Continental Shelf Project, Ottawa, Ontario; Royal Canadian Mounted Police Marine Unit, Yellowknife; and Water Survey of Canada, Yellowknife. Many dedicated graduate students and technical support staff have been involved in the various field campaigns and our thanks go to all of them. The authors thank the two anonymous reviewers of this manuscript for their very helpful critiques and suggestions.
REFERENCES
Austin, J. A., and Colman S. M. , 2007: Lake Superior summer water temperatures are increasing more rapidly than regional air temperatures: A positive ice–albedo feedback. Geophys. Res. Lett., 34 .L06604, doi:10.1029/2006GL029021.
Blanken, P. D., and Coauthors, 2000: Eddy covariance measurements of evaporation from Great Slave Lake, Northwest Territories, Canada. Water Resour. Res., 36 , 1069–1078.
Blanken, P. D., Rouse W. R. , and Schertzer W. M. , 2003: The enhancement of evaporation from a large northern lake by the entrainment of warm, dry air. J. Hydrometeor., 4 , 680–693.
Blanken, P. D., Rouse W. R. , and Schertzer W. M. , 2008: The time-scales of evaporation from Great Slave Lake. Hydrologic Processes, Vol. 2, Cold Region Atmospheric and Hydrologic Studies: The Mackenzie GEWEX Experience, M.-K. Woo, Ed., Springer, 181–196.
Bussières, N., and Schertzer W. M. , 2003: The evolution of AVHRR-derived water temperatures over lakes in the Mackenzie Basin and hydrometeorological applications. J. Hydrometeor., 4 , 660–672.
Chang, K. M., and Fu Y. , 2002: Interdecadal variations in Northern Hemisphere winter storm track intensity. J. Climate, 15 , 642–658.
Duguay, C. R., Flato G. M. , Jeffries M. O. , Ménard P. , Morris K. , and Rouse W. R. , 2002: Ice cover variability on shallow lakes at high latitudes: Model simulations and observations. Hydrol. Processes, 17 , 3465–3483.
Duguay, C. R., Prowse T. D. , Bonsal B. R. , Brown R. D. , Lacroix M. P. , and Ménard P. , 2006: Recent trends in Canadian lake ice cover. Hydrol. Processes, 20 , 781–801.
Gibson, J. J., Prowse T. D. , and Peters D. L. , 2006: Hydroclimatic controls on water balance and water level variability in Great Slave Lake. Hydrol. Processes, 20 , 4155–4172.
Johnson, L., 1975: Physical and chemical characteristics of Great Bear Lake, Northwest Territories. J. Fish. Res. Bd. Can., 32 , 1971–1987.
Johnson, L., 1994: Great Bear Lake. The Book of Canadian Lakes, Monogr. Series, No. 3, Canadian Association on Water Quality, 549–559.
Leon, L. F., Lam D. C. L. , Schertzer W. M. , and Swayne D. , 2005: Lake and climate models linkage: A 3-D hydrodynamic contribution. Advances in Geoscience, Vol. 4, Academic Press, 57–62.
Ménard, P., Duguay C. R. , Flato G. M. , and Rouse W. R. , 2002: Simulation of ice phenology on Great Slave Lake, Northwest Territories, Canada. Hydrol. Processes, 16 , 3691–3706.
Oswald, C. J., and Rouse W. R. , 2004: Thermal characteristics and energy balance of various-size Canadian Shield lakes in the Mackenzie River Basin. J. Hydrometeor., 5 , 129–144.
Rouse, W. R., Oswald C. J. , Binyamin J. , Blanken P. D. , Schertzer W. M. , and Spence C. , 2003a: Interannual and seasonal variability of the surface energy balance and temperature of central Great Slave Lake. J. Hydrometeor., 4 , 720–730.
Rouse, W. R., and Coauthors, 2003b: Energy and water cycles in a high-latitude, north-flowing river system: Summary of results from the Mackenzie GEWEX study—Phase 1. Bull. Amer. Meteor. Soc., 84 , 73–87.
Rouse, W. R., Oswald C. J. , Binyamin J. , Spence C. , Schertzer W. M. , Blanken P. D. , Bussières N. , and Duguay C. R. , 2005: The role of northern lakes in a regional energy balance. J. Hydrometeor., 6 , 291–305.
Schertzer, W. M., Rouse W. R. , and Blanken P. D. , 2000: Cross-lake variation of physical limnological and climatological processes of Great Slave Lake. Phys. Geogr., 21 , 385–406.
Schertzer, W. M., Rouse W. R. , Blanken P. D. , and Walker A. E. , 2003: Above-lake meteorology, thermal response, heat content, and estimate of the bulk heat exchange of Great Slave Lake during CAGES (1998–99). J. Hydrometeor., 4 , 649–659.
Schertzer, W. M., Rouse W. R. , Blanken P. D. , Walker A. E. , Lam D. C. L. , and Leon L. , 2008: Interannual variability in heat and mass exchange and thermal components of Great Slave Lake. Hydrologic Processes, Vol. 2, Cold Region Atmospheric and Hydrologic Studies: The Mackenzie GEWEX Experience, M.-K. Woo, Ed., Springer, 197–220.
Serreze, M. C., and Francis J. A. , 2006: The Arctic amplification debate 2006. Climatic Change, 76 , 241–264.
Shuttleworth, W. J., Gash J. H. C. , Lloyd D. , McNeil D. D. , Moore C. J. , and Wallace J. S. , 1988: An integrated micrometeorological system for evaporation measurements. Agric. For. Meteor., 43 , 295–317.
Thompson, D. W. J., and Wallace J. M. , 2001: Regional climate impacts of the Northern Hemisphere annular mode. Science, 293 , 85–89. doi:10.1126/science.1058958.
Walker, A., Silis A. , Metcalf J. R. , Davey M. R. , Brown R. D. , and Goodison B. E. , 2000: Snow cover and lake ice determination in the MAGS region using passive microwave satellite and conventional data. Proc. Fifth Scientific Workshop for the Mackenzie GEWEX Study, Edmonton, AB, Canada, Natural Sciences and Engineering Research Council and Environment Canada, 39–42. [Available from MAGS Secretariat, NWRI, Saskatoon, SK, S7N 3H5, Canada.].
Map of the Mackenzie River basin showing the locations of Great Slave and Great Bear Lakes, Hay River, Yellowknife, Norman Wells, and Deline.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Bathymetry of GSL (after Schertzer et al. 2000) with the position of Inner Whaleback Island indicated, and GBL (adapted from Johnson 1994) with the position of Lionel Island indicated. Depth contours are in m.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Long-term average (LTA) monthly (May–Dec) and annual (Y) regional temperatures for GSL and GBL. Here, MY plots the temperatures during the measurement years, and OW represents the average temperature for the open-water period. The regional temperatures at GSL are represented by the bias-adjusted averages of Hay River near the south shore and Yellowknife positioned on the northeast shore of GSL (Fig. 1). The regional temperatures for GBL are represented by Norman Wells, which is positioned to the WNW of Deline and Lionel Island (Fig. 1).
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Incoming solar radiation for GSL as measured at Inner Whaleback Island compared with GBL as measured at Lionel Island for comparable seasonal measurement periods but different years. The data for GSL are averaged for the years 1998, 1999, 2000, and 2002, while those for GBL are averaged for the years 2004 and 2005 (Table 1). The plots present 2-day moving averages. Seasonal averages are GSL = 233 W m−2 and GBL = 207 W m−2. The vertical bar denotes the summer solstice, the arrow with a dot the average time of breakup on GSL, and the solid arrow the average time of breakup on GBL.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Differences in overlake temperatures and regional air temperatures for (a) GSL and (b) GBL. (c) Comparison of overlake temperatures for GSL and GBL. The data represent 2-day running means.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Comparative lake surface temperatures in 1999 as derived from AVHRR analysis and measurements. (a) GSL-C and GSL-F are calculated daily temperatures and a second-order polynomial fitted to these calculations for GSL, and IWI-C and IWI-F are comparable calculations for a 5-km2 grid centered on the buoy near Inner Whaleback Island (IWI). (b) GBL-C and GBL-F are calculated daily temperatures and a second-order polynomial fitted to these calculations for GBL, and LI-C and LI-F are comparable calculations for a 5-km2 grid centered on the buoy near Lionel Island. [Note: GBL-F and LI-F are nearly identical and averaged as GBL&Li-F in (c).] (c) GSL-M and GBL-M are the averaged surface water temperatures taken during the measurement years shown with GSL-F, IWI-F, and GBL&Li-F.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Average daily water temperatures (3-day running means) as measured at IWI on GSL and Lionel Island on GBL. Tw0 and Tw40 designate the temperatures at the surface and at 40 m, respectively. Turnover temp is the water temperature at which vertical overturning is initiated.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Comparative (top) breakup dates, (middle) freeze-up dates, and (bottom) lengths of open-water periods for (left) GSL and (right) GBL, 1979–2005. The bars with the gray fill represent the measurement years used for most of the data analysis in this paper, the dark stippled bars represent the average of all years in the time series, and the light stippled bars represent the average for the measurement years. Data are derived from passive microwave time series analysis.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Comparative midlake heat storage in GSL and GBL for the years of measurement (Table 2). Note that calculations for the beginning and end of the ice-free period, when no data are available, are based on second-order polynomial fits to the measured data. The computations apply to a 45-m-deep 1-m2 column of water and represent the amount of stored heat relative to 0°C.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Boundary layer gradients showing 5-day running means of (a) vertical temperature gradients dT/dz and (b) vertical vapor pressure gradients de/dz. Positive gradients represent inversion conditions and negative gradients represent lapse conditions.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
(a) Latent heat fluxes (Qe) for GSL and GBL. (b) Sensible heat fluxes (Qh) for GSL and GBL. The trend lines (GSL-Fit and GBL-Fit) employ third-order polynomials. The trend lines are anchored at the freezing end by the mean dates of final freeze-back (Fig. 8) when Qe and Qh are assumed to be zero.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Comparative midlake (a) latent (Qe), sensible (Qh), and total convective heat fluxes (Qe + Qh), and (b) Bowen ratios (Qh/Qe) for GSL and GBL during their respective open-water seasons.
Citation: Journal of Hydrometeorology 9, 6; 10.1175/2008JHM977.1
Physical and morphometric characteristics of GSL and GBL: DA is drainage area, SA is surface area of the lake, V is lake volume, OW is the average open-water period, TW0 is approximate surface temperature during the OW period, Zmax is maximum depth, Zmean is average depth, D is downstream discharge from the lake, RT is estimated residence time of lake waters, and DLW and DLS are day length in winter (DJF) and summer (JJA), respectively.
Measurement and averaging periods for data that are presented in this study.
Regional temperature calculations using data from Hay River (HR) and Yellowknife (YK), where Ta = [Ta(HR) + Ta(YK)]/2 are biased due to the proximity of the two meteorological stations to the GSL south and north coasts, respectively (Fig. 1). It is assumed that, during east and west winds, the differences between HR and YK, ΔTaEW = Ta(HR)EW − Ta(YK)EW, represent true regional differences that are uninfluenced by GSL. During winds from the north and south, the corresponding differences, ΔTaNS = Ta(HR)NS − Ta(YK)NS, contain potential biases. These biases are seasonally dependant. The north–south wind direction biases are calculated as TaTB = ΔTaNS–ΔTaEW. The overall temperature bias correction is TaTBC = (TaTB) × (WDFNS), where WDFNS is the combined wind direction frequency from the north and south. A positive TaTBC indicates that that GSL is creating a warm bias in regional temperature calculations and a negative indicates a cold bias. The bias-corrected regional temperature is given by TaCORR = Ta − TaTBC. All temperature values are in °C. The temperature bias calculations are extracted from the mean daily temperature data in which the wind blows from the same direction at HR and YK for the data period 1953–2002.
Statistics of (a) the open-water period and (b) dates of thawing and freezing for GSL and GBL. In (a) and (b), avg, max, and min refer to the average, longest, and shortest open-water years.
Ice thicknesses (m) for GSL and GBL as measured in the winters of 2003/04 and 2004/05 and the 2-yr-average ice depths and dates of maximum thicknesses as modeled by the Canadian Lake Ice Model.
Estimated potential errors in calculations and measurements during the open-water season. RMS comparison refers to the combined root-mean-square error when comparing two measurements between GSL and GBL; Ta and Ta(OL) refer to the regional and overlake air temperatures, respectively; Tw is water temperature; StMAX is the lake heat storage at maximum; and Qe and Qh are the latent and sensible heat fluxes, respectively.
Differences in overlake air temperatures [Ta(OL)] and regional air temperatures (Ta) during heating and cooling periods for the 231 days from 15 May to 31 Dec inclusive. Here, (FZ) denotes when the lakes were frozen early and late in the season and (TH) denotes when they were thawed (Fig. 8). Positive values of Ta(OL) − Ta indicate that the overlake temperatures were warmer than the regional temperatures and negative values indicate that the lakes were colder than the regional temperatures.
