An oscillating bottom boundary layer connects the littoral and pelagic regions of Lake Opeongo, Canada

The movement of a thermocline can drive strong benthic currents, which can transport nutrients from sediments into the water column via pore water advection or sediment resuspension. We report field observations of near-shore benthic velocities and offshore thermocline movements in Lake Opeongo; a medium-sized lake typical of the Canadian Shield. We find that during large thermocline deflections there are sustained currents >6 cm s 1 in the near-shore benthic layer. The mean current was 1.75 cm s 1 and the maximum current is 10.3 cm s . At our site, the net transport is offshore even though the thermocline oscillates up and down so that currents are sometimes upslope and inshore. We estimate the excursion length of a water parcel over the 31-day deployment period, and determine that the mean daily excursion length is 630 m, with the maximum value being 2 km offshore. Given that the south arm of Lake Opeongo is 6 km long and 0.6 km wide, the predicted excursion length of water implies that there is strong connectivity between the sediment in the littoral zone, and the metalimnetic waters offshore. As Lake Opeongo is oligotrophic, any nutrient pulses from the sediment will be quickly taken up by the plankton. doi: 10.2166/wqrjc.2012.039 s://iwaponline.com/wqrj/article-pdf/47/3-4/215/163523/215.pdf Melissa Anne Coman Mathew Graeme Wells (corresponding author) Department of Physical and Environmental Sciences, University of Toronto Scarborough, 1265 Military Trail, Toronto, ON, Canada M1C 1A4 E-mail: wells@utsc.utoronto.ca


INTRODUCTION
It is important to understand the exchange of water and dissolved nutrients between the benthic and pelagic zones of a lake so that food web dynamics can be correctly modelled.
One mechanism of nutrient addition to the water column originates from pore water advection into the benthic water (Kirillin et al. ).Pore water is the interstitial fluid in the lake sediment and therefore has a high concentration of nutrients (e.g.nitrogen and phosphorus).Pore water is advected into the benthic water via a suction action driven by strong benthic currents (Søndergaard et al. ; Kirillin et al. ; Basterretxea et al. ) or through direct resuspension of sediments (Gloor et al. ).A slower, but continuous, process that also adds nutrients into the water column is diffusion across the nutrient gradient which exists between the sediments and the water column (Levine & Schindler ; Sundbäck et al. ).
The fate of any pore water that has been moved into the benthic region relies on the water circulation of the lake.Bottom turbulence will enable pore water to be mixed into the bottom boundary layer of the lake and subsequent transport of this fluid into the pelagic zone will make the nutrients available to a wider range of organisms.Absence of a mechanism to mix the pore water into the boundary layer or lack of transport into the pelagic zone will severely limit the usefulness of any nutrients contained in the pore water.Therefore, a good understanding of the ability of boundary layer fluid to be transported into various regions of a lake will be essential in understanding this pathway of nutrient addition to a lake's food web.
The circulation of water in Lake Opeongo is dominated by wind-driven seiche events (Coman & Wells ).For a lake or reservoir with a fairly uniform cross section, surface wind forcing will set up a horizontal pressure gradient along the lake axis aligned with the wind direction.If the lake is stratified with a strong thermocline (as many medium-sized Canadian Shield lakes are during summer time), this pressure gradient is balanced by a tilting of the thermocline.In this case, the thermocline is pushed downwards at the downwind end of the basin, and is pushed upwards at the upwind end of the basin as shown in Figure 1(a).The cessation or reduction of this wind forcing reduces the strength of the horizontal pressure gradient and releases the thermocline from the tilted orientation so that it relaxes back to its equilibrium level, shown in Figure 1(b).During this process the surface wind forcing has added momentum to the lake system so that upon wind reduction the thermocline does not simply return to its equilibrium level but undergoes seiching (shown in Figures 1(c) and 1(d)) until all the energy injected by the wind forcing has dissipated, or until another wind-forcing event re-establishes a new thermocline tilt.This oscillation is the basin scale seiche and is the major response of surface wind forcing in Lake Opeongo (Coman & Wells ) and other lakes.The deflected thermocline extends to the lake boundaries, and the movements associated with seiching cause currents along the benthic inshore regions.When the thermocline is upwelling, the current direction along the slope will be upslope and when the thermocline is downwelling the current direction will be downslope in the inshore benthic region.When an inshore site is experiencing an upwelling, the local water temperature will be decreasing as hypolimnetic water moves up the slope.Similarly, during a downwelling, the site experiences warming temperatures as the hypolimnetic water departs and is replaced with epilimnion water, as shown in Figure 1(e).Therefore, under this simple representation of the consequences of surface wind forcing on mediumsized lakes, we expect downslope currents and warming (a) The wind forcing has deflected the thermocline so that there is upwelling at the upwind end.(b) After the wind ceases the thermocline returns to its level of neutral buoyancy.(c) The thermocline then continues to oscillate such that maximum downwelling occurs at the original upwind end (time ¼ T/2) after which the currents again reverse and in (d) the thermocline is again at equilibrium level.One seiche cycle is complete if upwelling occurs again at the upwind end of the basin when time ¼ T. (e) The relative depth of the offshore isotherm and the relative temperature at a fixed depth at the inshore site.
temperature in the inshore region during a downwelling event and upslope currents and cooling temperature during an upwelling event (Figure 1(e)).In addition, if we measure isotherms at a more offshore location we expect the motion of these isotherms to be consistent with those inshore.That is, an upwelling event is caused by a deflected thermocline so this motion will be evident in the isotherm record of an offshore site (as long as the offshore site is not in the centre of the thermocline's tilting axis).If the offshore site is between the centre point of the lake wind axis and the inshore site (open triangle in Figure 1(b)) then an upwelling event will be preceded by isotherms that decrease in depth and a downwelling event will be preceded by offshore isotherms that move deeper into the lake.
In Lake Opoengo, the basin scale seiche is forced by westerly winds, and has a period of about 12 hours.Coman & Wells () found that the maximum displacement of the thermocline due to these motions has a strong dependence on surface wind forcing, lake stratification and basin morphometry.The largest displacement they measured was 10.7 m while the mean value was just 1.4 m.With a slope at the inshore site of 1%, these values imply a washing zone length of 1 km for the maximum thermocline displacement and a length of 140 m for mean displacement.
The magnitude of the thermocline tilt can be parameterised in terms of Lake number.The Lake number is a dimensionless parameter that combines the wind forcing and lake stratification, as well as including a measure of basin morphometry.Field observations in Lake Opeongo by Coman & Wells () found that the maximum displacement of the thermocline is proportional to the inverse of the Lake number.This result will apply to any lake with a strong and sharp thermocline (i.e.thickness of thermocline is less than the epilimnion or hypolimnion thickness) and a tri- The magnitude of benthic turbulence in a stratified lake is also related to the magnitude of the seiching and hence the Lake number.For instance MacIntyre et al. () found enhanced boundary mixing following low Lake number events in Mono Lake.Coman & Wells () also found that low values of the Lake number are associated with occurrences of temperature inversions in the nearshore benthic region.In Lake Opeongo, Coman & Wells () also found that the slope of the benthos played a role in the amount of temperature inversions measured at near-shore sites, most likely due to breaking of high frequency internal waves generated on the internal seiche (Michallet & Ivey ; Horn et al. ).Hence after a strong wind event the most turbulent benthic regions of a lake are expected to be those with lower slope angles that are located near the depth of the thermocline.These shallow slopes will also be the benthic regions that have the greatest temperature variability.was conducted in a small lake with a fetch in primary wind direction of 700 m.Energised boundary mixing was driven by an internal seiche in response to a strong wind event.
Rhodamine dye was injected into the boundary layer to act as a tracer and subsequent dye concentration mapping measured the three-dimensional (3D) extent of the tracer.
After 1 day the intrusion generated by boundary mixing was 0.5-1 m thick and reached over 200 m offshore.
Transport from the sloping boundary to the interior may also occur due to asymmetries in the benthic flow field as the thermocline moves up and down the lakebed.
Nakayama & Imberger () showed that due to details of the stratified turbulence on a sloping boundary there would be a net offshore transport.In a lake there may also be a basin scale circulation set up by wind-driven forcing, whereby on average water is drawn in at the edges of the lake and exits in the central axis of the lake, similar to the wind-driven double gyre circulation pattern described in Rao & Murty ().
In this paper we will quantify the benthic currents that are driven by oscillations in the position of the thermocline.
Our study takes place in the same region of Lake Opeongo

METHODS
The research presented here was conducted in Lake Opeongo, Algonquin Provincial Park, Ontario, Canada (Figure 2) between 15 July and 15 August 2010.Lake Opeongo is a low nutrient, medium-sized lake typical of many Canadian Shield lakes.Lake Opeongo contains four basins with a total surface area of 58.6 km 2 .We instrumented the South Arm basin which is roughly orientated in an east-west direction with a maximum fetch of 7. and a response time of less than 3 s.Four of these thermistors were attached to a vertical pole and deployed at 0.3, 0.6, 0.9, and 1.2 m above the sediments in 7 m of water.
The temperature was recorded every 4 s.The temporal drift over the 31-day deployment was at most 8 s so no adjustment was made to the timing recorded by the thermistor units.A 3 beam, 1.5 MHz Acoustic Doppler Profiler (SonTek, San Diego, CA) was deployed on an A-frame facing downwards at the inshore site.This instrument was able to measure the velocity in the bottom 60 cm of the 7 m water column (from 1.5 cm above the sediments).The cell size was 3.1 cm with a blanking distance of 5 cm.
Burst mode was used so that 24 profiles were collected every other minute.For the purposes of this research five of these 1 minute bursts were averaged to leave a data set with a 10 minute resolution.
The eastern, predominately downwind end of the South Arm basin of Lake Opeongo is covered with large The distance a parcel of water is advected away from the boundary can be estimated if the benthic velocity field is known.The excursion length during a given time period, Δt, is equal to the average velocity over that time period multiplied by the time period, Since the current over Δt may be changing magnitude as well as direction we use circular statistics to find the average velocity, u over any period, Δt.The summation of the excursion lengths of individual time periods (Equation (1)) will be equal to the integral of the velocity over time, where u and v are the speeds in the x and y directions, respectively, t s is the initial time (July 15 in our case) and

RESULTS
The benthic currents are correlated to changes in the stratification caused by upwelling or downwelling events at the western end of Lake Opeongo.A new observation is that the magnitude of the benthic offshore currents is greater than the benthic onshore currents.
Rather than a simple oscillation up and down the slope, there is a net offshore transport of water from the benthic inshore region of the lake.This observation warrants further study to determine whether there is either an asymmetry  current events in our record occur when the inshore temperatures are increasing, that is, during the of an upwelling event (caused by westerly winds) or during the beginning of a downwelling event (caused by easterly winds).
The estimates of the large excursion lengths of the water parcels suggest that there should be strong exchange between the littoral and pelagic zones of Lake Opeongo.
The measurements of strong unidirectional benthic currents that last for hours (rather than minutes) gives rise to the potential for substantive offshore water transport in Lake Opeongo rather than localised transport contained to the inshore region due to evenly oscillating currents.
The depth at which water transported from the inshore region will enter the larger lake basin will depend upon the density of the mixed inshore benthic water and of the stratification offshore.The downslope and faster currents typically occur as the upper portion of the thermocline is moving down the slope.During the sampling period the thermocline was located close to the 15 W C isotherm at depths between 7 and 10 m.Hence water from the littoral zone that is potentially nutrient rich, will enter the pelagic waters of Lake Opeongo in the upper thermocline, where it may subsequently be available for primary production.
The speeds reached during the large downslope currents were often greater than 6 cm s À1 so are large enough at the inshore site to draw pore water and associated nutrients out of the sediments ( within the 1-5 m depth range as occurs at 7 m depth then nutrients resuspended via surface waves may also be transported offshore, perhaps at shallower depths given the warmer temperatures that will be present at sites further inshore of our inshore site.

CONCLUSIONS
In Lake Opeongo, the wind-driven internal seiche drives benthic velocities in the near-shore regions that are on average 1.75 cm s À1 , with brief periods where the maximum current is up to 10 cm s À1 .The current is greater than 6 cm s À1 on several occasions and in all of these occurrences the current direction is offshore, that is, flowing down the slope.In general we find that the downslope currents are faster than the upslope currents at our inshore site, located at the upwind end of South Arm basin.Near bed velocities greater than 6 cm s À1 , occur only 4% of the time, suggesting that there is infrequent potential for sediment resuspension.The maximum mean daily transport occurs for day 199.5-200.5 in an easterly (that is, offshore) direction, and had a value of 2 km per day.The average daily mean transport over the deployment period is 630 m in a south east direction (134 N).
Transport of bottom water occurs most strongly in the offshore direction when isotherm motions due to thermocline deflection are largest.Calm periods show either an equal onshore/offshore transport or a small net offshore transport.While the mean currents are unlikely to resuspend any sediment, they can result in daily advection of water of 0.6 km.As the South Arm of Lake Opeongo has a length of 6 km and a mean width of 0.6 km, the advection by the mean current means that the water at the depth of the thermocline in the near-shore zone is able to exchange with the pelagic waters.This constant exchange has implications for the nutrient cycling in the water column, and suggests that near-shore and offshore sites are well connected, particularly after large seiching events.
We found that the near bed velocity at 7 m depth was greatest after strong seiche events, and on average was directed offshore.Such an asymmetry in the near bed velocity could be due to processes associated with boundary mixing described by Nakayama & Imberger (), or could indicate a more complicated 3D circulation pattern in the lake in to wind-driven forcing.The velocity exceeded 6 cm s À1 about 4% of the time.This occurred during seven strong events (day 198, 199, 200, 209, 211, 217 and 218), which were all associated with corresponding seiche driven offshore isotherm and inshore temperature movements.We expect these events in particular increased nutrient addition to the benthic boundary layer through advection of pore water and possibly occasional (<0.7% of the time) sediment resuspension given speeds >7 cm s À1 that were occasionally reached (Gloor et al. ).The mean current speed is expected to enable pore water advection along the mean excursion length, which can still connect the near-shore and offshore regions at the depth of the thermocline.Therefore any nutrients pumped into the water column have the potential to be transported far offshore under strong wind conditions or at least into the pelagic zone under mean wind conditions.Given that Lake Opeongo is oligotrophic, nutrients are likely absorbed quickly by local plankton which themselves can then be transported into the pelagic zone to become available as food to other trophic levels.

Figure 1 |
Figure 1 | Schematic of thermocline tilting and inshore current direction.The solid line represents a thermocline, and the direction of currents and water movements are shown by arrows.Solid triangle represents relative inshore site location and open triangle represents relative offshore site location.Time is shown as a fraction of the basin scale seiche period, T.
angular basin morphometry.This relationship was initially partially confirmed by Stevens & Lawrence () who measured the Wedderburn number and thermocline deflection in four different sized Canadian lakes.
There are many observations of large-scale internal seiches in lakes (Lorke ; Wells & Parker ).Resuspension of bottom sediments due to lake seiches has been directly observed (Shteinman et al. ) by tracking marked particles in the bed of Lake Kinneret, resulting in resuspended phosphates which enhanced the algal productivity in the water column (Ostrovsky et al. ).Indirect observations of particle resuspension due to large amplitude lake seiches include observations of elevated suspended particle concentrations near the lakebed (Gloor et al. ; Hawley & Muzzi ) and observations of increased phosphate fluxes after strong seiche events (MacIntyre et al. ).The resuspended phosphates arising from benthic turbulence may either be used locally or may be taken offshore by horizontal intrusions (Wain & Rehmann ) formed on the sloping boundaries of the lake.Several field experiments have documented the role of benthic turbulence in driving intrusions away from a sloping boundary.Inall () conducted the first field experiment that measured the fate of dye injected just above the sloping boundary layer in a long fjord where boundary mixing occurs via the interaction between the semidiurnal tide and rough topography.The dye initially spread in the vertical, a portion entered the boundary layer and subsequently intruded into the interior of the fjord in three distinct layers.The intrusions were driven by the gravitational collapse of the mixed boundary layer fluid and subsequent intrusion along isopycnal surfaces.The experiments of Wain & Rehmann () tracked an intrusion from dye injected directly into the turbulent boundary layer on a sloping boundary (5-10%) into the interior of a lake.Their study as described byComan & Wells ()  and extends this study with new observations of the benthic currents measured with a downwards facing acoustic Doppler profiler.The main focus of the paper is to relate the observations of benthic currents and changes in water temperature to the occurrence of seiching motions in the lake.Specifically, we show that oscillating currents lead to a net transport of water from the shallow littoral zone to the offshore pelagic zone.
Figure 2 | Map of (a) Lake Opeongo, and (b) the South Arm of Lake Opeongo with the inshore and offshore sites indicated with a triangle and circle, respectively, and the weather station shown by the star.The contour interval is 5 m.
rocks and boulders down to about 5-6 m.A fine layer of sediments settles on these boulders (Finlay et al. ; McCabe & Cyr ).At depths greater than 5 m, soft substrates are dominant.More sheltered subsections of the eastern shoreline are characterised by deep sandy substrates below 1 m depth.The western end of the basin is predominately upwind of the prevailing winds and is therefore generally more sheltered compared to the eastern end.The substrate in the west is typically coarse sand and small rocks down to 0.9 m with a soft substrate of sand mixed with mud covering the bedrock below about 1 m (Finlay et al. ; McCabe & Cyr ).Below 7-8 m, the lake sediment is dominated by gyttja, and the sedimentation is not influenced by surface waves except during large storms in the fall (Finlay et al. ; McCabe & Cyr ).Nitrogen and phosphorus are both available in the sediments or pore water of lakes (Søndergaard et al. ; Kirillin et al. ).The potential for phosphorus release during sediment resuspension events in near-shore areas of Lake Opeongo was shown by Cyr et al. ().Given the oligotrophic nature of Lake Opeongo, any nutrients that are released from the sediments (via any method) will contribute to the available food source for phytoplankton.The differences in substrate among different near-shore regions of Lake Opeongo (mudcoarse sandsboulders covered with fine sediments) may mean the ratio of nitrogen to phosphorus released from pore waters will differ among regions (McCabe & Cyr ; Cyr et al. ).
t f is the final time (August 15).To calculate Equation (2) we first depth average the velocity data then calculate a running 10 minute temporal average of the velocity field.From this smoothed data the u and v components are taken to be along the vectors parallel and perpendicular to the major South Arm basin fetch, i.e. at 100 W and 190 W from N. Therefore L(t) off is the average advection directed offshore, and L (t) long is the advection directed alongshore.We are neglecting vertical velocities in our estimates of advection lengths, as the observed vertical velocities are generally close to the detection limit of our Acoustic Doppler Profiler.These estimates of the Lagrangian particle excursion lengths L(t) are based upon Eulerian measurements, so do not follow a particular 60 cm deep benthic packet of water from the inshore site out into the pelagic region of South Arm as the horizontal velocity field is not uniform throughout the lake.Rather the particle excursion measure, L(t) off , gives us an idea of the potential for water transport originating from this inshore benthic region.If anything, L(t) is likely to be an underestimate of the advection scale of these water masses as the horizontal velocities induced by a basin scale seiche increase in the offshore direction.For example in a simple rectangular basin, the first horizontal mode wave has a sinusoidal form with the horizontal velocity maximum in the middle of the lake (Fricker & Nepf ).
Figure 3 | (a) Current speed at the inshore site, (b) current direction at the inshore site, (c) isotherms at the offshore site with the 15.5 W C isotherm in blue, 17.5 W C in green and 19 W C in red, and (d) isodepths (temperature at a fixed depth) at the inshore site, 0.3 m (black), 0.6 m (blue), 0.9 m (green) and 1.2 m (red) above the sediment.The full colour version of this figure can be found online at http://www.iwaponline.com/wqrjc/toc.htm.

Figure 4 Figure 5
Figure 4 | (a) A current rose showing the direction of benthic currents.(b) Total advection distance (km) over the 31-day deployment period as a function of direction at the inshore site.North is 0 W .
Figure 6 | (a) Mean daily excursion length (km), and (b) the direction of excursion.

Figure 7 |
Figure 7 | Normalised histogram of current speeds from the inshore site over the deployment period.These speeds are 1 minute averages from the 60 cm range above the sediment.The mean current speed of 1.75 cm s À1 is shown as the vertical dashed line.