4

In-Lake Processes: Hydrodynamics

The goal of the next three chapters is to analyze Coeur d’Alene (CDA) Lake water column data collected over the past 30 years in order to assess trends in water quality. In conducting its analyses, the committee set about to elucidate the main water column processes at work affecting water quality in order to make estimates about future conditions, which are elaborated on in Chapter 10. Unlike Chapter 3, which focused on metals and phosphorus, the in-lake analyses consider additional parameters relevant to the condition of the Lake, including physical parameters such as water motions, specific conductance, pH, and temperature (Chapter 4); dissolved oxygen, nutrients, and other parameters indicative of the Lake’s productivity, such as chlorophyll a (Chapter 5); and metals, including arsenic (Chapter 6). In addition to illuminating water quality trends, the available in-lake spatial and temporal data are evaluated by the committee to better understand the Lake’s potential for eutrophication and, ultimately, metals mobilization from sediments, which is the subject of Chapter 7.

INTRODUCTION TO LAKE PROCESSES

Seasonal and long-term water quality trends within a lake are the result of the interactions among the key physical, chemical, and biological processes and the associated process drivers that alter inputs, outputs and internal dynamics within the lake, as illustrated in Figure 4-1. The main high-level processes relevant within CDA Lake include the lake hydrodynamics, sediment transport, sediment biogeochemistry, and primary productivity (examples of specific processes within those four categories are given below). Each of these processes is influenced by a subset of drivers that can affect the water quality within a lake both spatially and temporally. These drivers include riverine inputs (e.g., metal, nutrient, and sediment loads) associated with river flows and lake inputs, such as the meteorological forcing at the air–water interface. The drivers determine the mixing, chemical loads, and photochemical and biogeochemical dynamics within a lake that lead to changes to water quality parameters, such as pH, phosphorus and nitrogen concentrations, dissolved oxygen, and chlorophyll a.

Figure 4-1 illustrates a number of the key interactions among processes and drivers in CDA Lake and is referred to repeatedly in the next four chapters. This chapter covers relevant hydrodynamic processes, such as the upstream discharges from the St. Joe River and CDA River. These inflows enter CDA Lake at variable depths, depending on both river and lake temperatures (which change seasonally), and they contribute to the general south-to-north advection within the Lake. The CDA River provides the main source of particulate and dissolved metals, while the cleaner St. Joe River serves to dilute metal concentrations in the southern reach of the lake (e.g., at locations such as C5), with the extent of dilution controlled by in-lake hydrodynamics and physical processes

such as sediment transport. Internal hydrodynamics have redistributed particulate metal contamination through the northern and southern Lake, as evidenced by widespread sediment contamination. The details of these hydrodynamic processes are not well known (as discussed below) but could be critical to understanding recovery of Lake water quality into the future.

Also discussed in this chapter are sediment transport processes that can have a significant impact on water quality within CDA Lake. Turbidity in the water column impacts photosynthesis by reducing the depth of light penetration. Settling of particles including inorganic solids and algae detritus can transport metals and nutrients from the euphotic zone of the water column to bottom waters and the sediment. If remediation reduces the particulate metal load to the Lake, then particles entering the Lake that settle to the bottom can serve as a capping layer over previously deposited sediments.

Discharges from the rivers as well as nonpoint sources can influence phosphorus concentrations in the Lake, which may lead to increases in phytoplankton biomass and production in the Lake as lake temperatures warm and the days get longer in the spring and summer. Dissolved and particulate organic matter in the water column can bind metal ions and alter their concentrations. A yearly occurrence in lakes (including CDA Lake) is thermal stratification, in which colder and denser bottom waters become isolated from warmer surface waters. Thermal stratification allows some dissolved metals to accumulate in bottom waters, impacts sediment transport through reduced velocities, and can lead to lower dissolved oxygen concentrations in the lake bottom waters (hypolimnion).

Such in-lake processes affecting dissolved oxygen, nutrients, and chlorophyll a are explored in Chapter 5, while the processes affecting metals are discussed in Chapter 6.

At work in the bottom waters of CDA Lake are sediment biogeochemistry processes that control metal speciation, adsorption/desorption to sediments, and pore water solubility through changes in pH and redox conditions, as well as lake hydrodynamics like mixing that control the inputs and exchanges to the sediment. In addition, reduced pH caused by organic particle decomposition in low dissolved oxygen areas of the water column (Davison, 1993) can lead to metal ion (e.g., zinc) release from settling particles and from bottom sediments into bottom waters, as observed by Kuwabara et al. (2000). These processes are all discussed in Chapter 7.

Figure 4-1 is representative of in-lake processes during the summer. The drivers that influence metal concentrations and speciation, thermal stratification, nutrient cycling, and primary and secondary productivity all change seasonally. For example, spring snowmelt leads to higher river flow rates and the associated increase in sediment transport. Particles entering the Lake in the spring are primarily inorganic materials from the watershed, including colloidal/nanosized materials (see, e.g., Davison, 1993). When inflows decrease in late spring and early summer, inorganic materials settle in the Lake or are exported via the Spokane River, the Lake clears, and productivity increases. Particulate material becomes dominated first by phytoplankton and then by detrital organic particles through summer and fall. In some lakes, the settling of organic particles during and after periods of phytoplankton growth can strip some metals from the water column and rerelease them in deeper waters (Balistrieri et al., 2002). Seasonal changes in dissolved oxygen and pH due to photosynthesis can also have significant impacts on metal speciation and the abundances and diversity of biota in the Lake—both at the sediment–water interface and within the water column itself. Higher water column oxygen concentrations in the winter and spring result from the higher oxygen solubility at lower temperatures, but phytoplankton productivity in the summer can further increase dissolved oxygen concentrations in the epilimnion. Throughout this and the two subsequent chapters, the available data are analyzed to assess the impact of seasonality on metal and nutrient concentrations.

LAKE HYDRODYNAMICS

The hydrodynamics of a lake involve the water movements that arise on account of external forcing (such as the surface heat exchange, wind, inflows and outflows) and internal responses to that forcing. Both occur within a physical domain (the system boundaries). In the case of the external forcing, that physical domain is the watershed (described by the topography and land use types), while for a lake it is the lake basin, defined by its bathymetry and the water surface.

The external forcing is impacted by the surrounding topography; the spatial characteristics of the watershed; and the temporal changes in the hydrology, meteorology, and climate. These temporal changes occur over a broad range of scales (minutes through years). Alterations in land use within the watershed can also exert impacts. For example, after land clearing, enhanced erosion can increase the suspended sediment concentration of a stream, and hence its nutrient load and, in extreme cases, its density relative to the lake water. These changes generally occur at longer timescales than variability in meteorology and hydrology.

The responses of the lake to the forcing are numerous and complex, particularly at those times of year when a lake’s water column is thermally stratified (and hence density stratified). When a lake is thermally stratified, physical forces bring it to a state in which isotherms (contours of equal temperature) are horizontal, thereby preserving a minimal potential energy state. When external forcing is imposed, such as by a wind blowing over the surface, the isotherms become tilted, and lake motions are initiated to reestablish horizontal isotherms. Such motions are characteristically oscillatory and are often referred to as internal waves. Stratification generally occurs during the warmer times of year (spring, summer, fall), but even winter can exhibit weak stratification that exerts an influence on the hydrodynamics. This is even true for times of the year when a lake or part of it is ice-covered (see Wüest and Lorke, 2003).

Figure 4-2 shows a simplified case of a two-layer stratified lake to illustrate how a lake responds to external forcing. The responses are of two general types: barotropic and baroclinic. Barotropic responses are exemplified by the action of the wind driving large, basin scale, horizontal gyres of the surface waters of the lake. The surface drag force allows momentum from the wind to be transferred to the water, and the gyres are constrained by the

shape of the lake basin. The wind can also cause sloshing motions of the water within the basin, known as surface seiches. These basin-scale surface waves typically have periods on the order of minutes for lakes the size of CDA Lake, and amplitudes of millimeters (see Figure 4-2, where the surface seiches are shown to have an amplitude of ζ_o). More apparent are wind waves that move across the lake surface. These typically have periods of 2 to 5 seconds and maximum amplitudes of up to 1 to 2 meters. Though these are often the most visible lake motion, they are generally of little importance outside of the littoral zone, where they can break and cause shoreline erosion and resuspend sediments. For lakes the size of CDA Lake, the earth’s rotation has no effect on barotropic motions.

Baroclinic motions arise when horizontal pressure gradients are established, due to tilting of a lake’s isotherms, as described previously. Water of two different temperatures at the same horizontal level across the lake (if, for example, the lake isotherms were tilted) would constitute a horizontal pressure gradient. This arrangement is gravitationally unstable, and internal lake motions ensue that eliminate such gradients. The motions take the form of internal waves and vertical upwellings (water rising up at the boundary) and downwellings (water forced down near the boundary). In Figure 4-2, the internal wave amplitude is shown as ζ₁(x). On the left side of the schematic, where the interface is deviated upward, upwelling is occurring, while on the opposite side there is downwelling. Once the wind stops, the internal wave oscillates back and forth until it is damped by viscosity. The effect of the earth’s rotation complicates these motions further, and phenomena such as Kelvin waves (edge-trapped waves) can also arise. The baroclinically driven motions occur over long timescales (hours to days), can generate high current velocities (tens of centimeters per second), and often exhibit large vertical amplitudes (meters to tens of meters).

The hydrodynamic responses of a lake to external forcings can be broken down into a number of individual processes that will now be briefly described. As the responses depend to a great degree on whether a lake is stratified or not, the discussion of processes is separated into “summer” and “winter” to denote the strongly stratified (summer) conditions and the well-mixed or weakly stratified conditions (winter).

Summer Hydrodynamics

During summer in temperate latitudes (see Figure 4-1), lakes stratify in temperature due to a net positive heat exchange at the surface (more heat gained than lost). Surface mixing processes due to wind mixing, and convective cooling of the surface result in the formation of a surface mixed layer (epilimnion) separated from the deeper water by a sharp gradient in temperature (the thermocline). The underlying water is referred to as the hypolimnion, and it is often weakly stratified. Stratification is typically a continuum, with the strongest stratification at the thermocline and the weaker stratification below. An important feature of stratification is that it inhibits

the vertical exchange of dissolved substances, including nutrients, metals, and dissolved oxygen. The latter is particularly important because in deep portions of a lake below the thermocline, dissolved oxygen is generally being lost due to decomposition and sediment oxygen demand. After sufficient time, this loss can lead to hypoxia (low oxygen) and eventually anoxia (no oxygen)—conditions that favor the release of some nutrients, metals, and metalloids from the sediments.

Inflows from the streams arrive with a temperature that determines their density. In extreme cases, high suspended sediment concentrations can increase that density. When a stream inflow reaches a stratified lake, its density can be either lower than, higher than, or the same as the lake’s. Depending on which case it is, the stream will either overflow, underflow, or form an interflow near the depth of the thermocline. Whichever it does, the inflow carries its “load” of constituents and these get inserted at that depth. If the loads are particulate, they will in time settle out. Note that the stream temperature and the lake surface temperature are both changing throughout the day, often with a significant time lag for the air temperature. Thus, the relative density difference between the lake and the inflow can vary throughout the day and throughout the season. The flowrate and the shape of the inflow channel also play a part in enhancing the mixing and dilution that occurs to the inflow as it is entering the lake. The dilution of the particulate and dissolved contaminant concentrations in the river inflow can sometimes be as high as a factor of 10 (Ayala et al., 2014).

Bottom mixing processes are driven by currents moving across the lake bed. Generally, hypolimnetic currents in a lake are weak (on the order of 1 cm/s), resulting in relatively little mixing. However, higher velocities can be generated by baroclinic motions and internal waves (10–50 cm/s), and these can result in mixing. They can also result in sediment resuspension (Roberts et al., 2021) and reoxygenation of the sediments.

Basin-scale gyres (see Figure 4-3) at the surface of a lake are generally the result of the wind blowing across the lake surface. Though playing little role in the vertical mixing process of the lake, they can distribute epilimnetic material very efficiently and could, for example, trap algal blooms in a particular embayment.

In winter, the same range of hydrodynamic processes are at play, but with some important differences. Often the lake temperature is isothermal due to lake cooling processes that keep the water body well mixed. As a result, dissolved and suspended material are generally isotropic (no direction gradients). Weak stratification can sometimes persist. If ice forms, a reverse temperature gradient can be established that will stratify the lake, with the heaviest 4°C water at the lake bottom, and lighter 0°C water immediately below the ice. Under these conditions, dissolved oxygen will again be lost at the bottom (although decomposition slows at the lower temperature), and it is possible for hypoxia to occur at the lake bottom. With the physical ice cap on the lake, surface aeration is cut off, and dissolved oxygen can be depleted at a faster rate than in summer. Typically, inflows are low during winter, so while the same range of processes can all occur, they generally have a small effect on a lake.

Hydrodynamics Data Analysis

There have been relatively few physical limnology and hydrodynamic studies of CDA Lake, beyond the long-term monitoring conducted by state and tribal entities (see Chapter 2). This has led to a paucity of information on the key physical processes at play in the lake, and a near total absence of time-series data that would allow such an understanding to be developed. The two exceptions are (1) a detailed 12-day field measurement program conducted in 2005 and associated three-dimensional (3-D) model development completed in 2007 (Dallimore et al., 2007; Hipsey et al., 2007; Morillo et al., 2008) and (2) the deployment of a continuous profiling station by the CDA Tribe at station C5 during three periods (in 2011, 2015, and 2019).

2005 Experiment and Modeling

The 2005 experiment involved the installation of a thermistor chain and weather station called LDS, repeated transects along the thalweg of the Lake (plus several cross-sections) using a multi-parameter probe (F-Probe), and microstructure profiling at one station (see Figure 4-4). The experiment ran from approximately June 1 to June 11, 2005, and coincided with a time when the St. Joe River at Chatcolet had flow rates falling from 3,590 cfs (112 cms) to 2,350 cfs (67 cms), and the CDA River at Harrison had flow rates falling from 1,860 to 1,260 cfs.

These flowrates are mentioned specifically because their magnitudes appear to be an important determinant in the response of the Lake throughout the subsequent summer months.

Some representative temperature transects from the experiment are shown in Figure 4-5A. The mouth of the St. Joe River is to the left on the transects, and the mouth of the CDA River is at station 13. Station 60 (far right) is the main basin of CDA Lake opposite the city of Coeur d’Alene. The transects are composed of a number of F-Probe vertical profiles (locations marked by upside down triangles), and each transect took several hours to complete.

The vertical positions of the river inflows can be inferred from the lowest salinity (lowest specific conductivity) water in lake profiles, as both rivers are relatively low in conductivity compared to the Lake. In Figure 4-5B, which shows the corresponding transects for salinity, the dark blue region of low conductivity at the left on Panel 2 shows that the river water from the St. Joe entered the Lake as an interflow, meaning that because of its density relative to the stratification of the Lake, it intruded below the thermocline but above the lake bottom. The bright

red region of high conductivity on the third panel shows an overflow being produced by the CDA River, intruding into the epilimnion above the thermocline. These observations suggest that the river inflows are not likely to play

a major role in resuspending sediment within the Lake under the conditions shown. The low Froude numbers of the inflows (< 0.21), a metric that expresses the magnitude of the inflow momentum relative to the strength of the vertical stratification, also suggested that the intrusions did not possess sufficient momentum at this time to produce significant mixing, and that any water column mixing would either come from internal wave motions or from external wind-driven processes. Clearly, this may be different under higher flow conditions, although the flows would likely have to increase by a factor of 5 for this to occur.

Internal wave motions (not shown) were evident in the thermistor chain data (collected at 60-second intervals). Spectral analysis (a mathematical technique that allows identification of dominant, periodic motions) did not identify any obvious peaks, possibly because of the short observation period (11 days). There was little to suggest that internal wave (baroclinic) motions would be a major factor in CDA Lake mixing, although the data are very limited.

The ELCOM hydrodynamic model (see subsequent section), which underwent limited calibration and validation using the short data record from this experiment, was run to examine the transport between the lower east–west portion of the Lake, and the upper north–south portion of the Lake. The field data and the model results indicated that the dominant horizontal seiches had the effect of temporarily trapping river water in the lower part of the lake, in the region of the two river inflows. This observation may in part explain subsequent observations of high suspended metals concentrations at station C5 (see Chapter 6). Numerical “tracer” experiments were also conducted

with the model, whereby modeled tracers were released from the St. Joe and the CDA Rivers. The model results suggested that the tracers from both rivers quickly became distributed through the epilimnion, as the observations suggest. It should be noted that the model was not run to explore conditions outside those observed during the experimental period or to infer what may be happening later in the summer or during the winter months.

Continuous Profiling at C5

CDA Tribe data from continuous measurements at site C5 were collected over the summer and fall of three different years, and therefore offer insights into the Lake response under three different sets of hydrologic conditions. While some of the observations that were made in the 2005 experiment are confirmed, the data suggest that the magnitude and timing of each year’s river flow is a key variable in both the physical limnology and the water quality response. Furthermore, it should be remembered that there is a distinct difference in both the physical and biogeochemical behavior of the northern and southern (south of the mouth of the CDA River) Lake, such that the C5 data are not representative of the northern part of the Lake.

Figure 4-6 presents annual (water year) hydrographs for the St. Joe River for 2010–2020. Although each year is different, a notable characteristic that each year displays is the steep, monotonic falling limb, commencing around June. During high flows earlier in the year, the lower part of CDA Lake where station C5 is located is dominated by the flow from the St. Joe River, and it is only when the flow has fallen low enough that the Lake stratification starts to dominate the hydrodynamics and the water quality, particularly in the hypolimnion.

Station C5 (see Figure 4-4) was the location of the continuous profiling instrument that was operated by the CDA Tribe for several months in 2011, 2015, and 2019. In 2011 and 2015, data on temperature, electrical conductivity, pH, and dissolved oxygen were collected. In 2019, only temperature and electrical conductivity data were collected. The C5 profiler time series of temperature, specific conductivity and dissolved oxygen from 2011 are shown in Figure 4-7. The red bars at the top of the figure indicate when data were available (there are several data gaps that have been filled by interpolation). The flow for the St. Joe River between June 1, 2011, and December 31, 2011, is shown at the bottom.

As can be seen in Figure 4-7, temperature and specific conductance both indicate the presence of vertical oscillations throughout the data.¹ The temperature data indicate that thermal stratification is very surficial until early August, but then increases in intensity, and the depth of the thermocline descends from about 5 m to 8 m before cooling commences in early September. Stratification persists through mid-October. The specific conductance figure (middle of Figure 4-7) shows the clear signal of the St. Joe River intrusion as the blue “tongue.” At the beginning of the plot, the dark blue color indicates that the water from the St. Joe is entering at the surface, but as the flow rate reduces over the summer, the intrusion is more dilute (it has entrained lake water) and is located lower in the water column but above the bottom of the Lake. The lowest panel shows dissolved oxygen concentration; low oxygen forms at the bottom of the Lake, but only falls to about 5–6 mg/L. The transition from inflow-dominated conditions to stratification-dominated conditions would appear to commence in early August, when the river flow was in the vicinity of 1,000 cfs.

The year 2015 was very different, particularly for stream inflow (Figure 4-8). At the beginning of the recorded period at C5, stratification was already established, with surface temperatures in excess of 20°C. Unlike 2011, there was no inflow-dominated period evident, as river flows were already below 800 cfs when profiling at C5 commenced. The specific conductivity indicated that the river inflow was present as a distinct underflow (dark blue in Figure 4-8), one that persisted throughout the year until vertical mixing occurred in late November. Most notably, dissolved oxygen became depleted at the bottom of the hypolimnion in mid-July and remained so until full vertical mixing occurred. The shape of the dissolved oxygen distribution suggests that the cause of the dissolved oxygen decline was sediment bed–induced, rather than due to biological oxygen demand from the river inflow. At times, dissolved oxygen levels were below 2 mg/L. It is not known how close to the sediment–water interface the instrument profiled, but it is believed to be approximately 1 m off the bottom.

___________________

¹ Although the zigzags represent internal waves, the committee does not interpret these waves to mean that there is high-intensity turbulence in the Lake that produces a lot of mixing. Waves are very efficient at transporting energy and produce mixing only when they break.

Conditions for the deployment of the C5 profiler in 2019 were intermediate between the two earlier deployments. The deployment was shorter (three months) and only temperature and specific conductivity were measured (Figure 4-9). Based on the position of the conductivity minimum, the river intrusion occurred as an interflow. Unfortunately, the absence of dissolved oxygen measurements precludes a better understanding of the impacts of the interflow. As was the case in 2015, the river flow fell below 1,000 cfs prior to commencement of profiling, so the transition between river-dominated and buoyancy-dominated conditions could not be observed.

Plotting the St. Joe River inflow against dissolved oxygen and pH at near-bottom (16 m depth) at station C5 reveals a relationship between river flow and water quality. In Figure 4-10, the 2011 profiler data from midnight on

each day have been plotted (profiling actually took place at two-hour intervals). The year 2011 was a large water flow year, such that the St. Joe River did not fall below 1,000 cfs until early August. Figure 4-10 clearly shows that water quality (dissolved oxygen and pH) does not change systematically when flows are above 1,000 cfs and the system is in the inflow-dominated mode. Below this flow, when the system is transitioning to the buoyancy-dominated mode, both variables decrease in near-linear fashion, with dissolved oxygen falling from approximately 9 mg/L to 5 mg/L and pH falling from approximately 6.9 to 6.5. That is, below 1,000 cfs, the system transitions from river-dominated to the summer stratification period; this may happen in June, July, or August.

In the low-flow year 2015, the St. Joe River was flowing well below 1,000 cfs when the profiler at C5 was installed. Looking at an equivalent plot for 2015 (Figure 4-11), the data all lie along the falling trend-line seen in Figure 4-10, with dissolved oxygen decreasing from 9 mg/L to below 4 mg/L and pH falling in the range of 6.9 to below 6.4.

It is noteworthy that the 2005 hydrodynamic experiment (see previous section), probably the most extensive set of hydrodynamic measurements existing for CDA Lake, was made over a ten-day period when the flow in the St. Joe River was dropping from 3,590 cfs to 2,350 cfs. In other words, the upper portion of the Lake (in the vicinity of C5) where the largest water quality impacts are currently observed was in the midst of the inflow-dominated regime, and the impacts of thermal stratification were likely not occurring and therefore were not captured.

The take-home message from these analyses of hydrodynamic data is that “lake-like” conditions in CDA Lake only occur at low St. Joe River flows (defined as less than 1,000 cfs). Currently at C5, the length of this period is variable, such that hypoxic conditions start to form in some years where 1,000 cfs is reached sooner. If earlier river peaks were to occur under climate change, this period of lake-like conditions could commence sooner, extending the time for oxygen loss. Likewise, if warmer air temperatures in summer due to climate change protracted the stratified season (as it has done in many lakes around the world), then the stratification season is lengthening even further. This would be pushing the lower Lake (at C5) toward anoxia, and may create conditions in the northern Lake for hypoxic episodes in the future. The committee cautions that the C5 data are not representative of the northern Lake; thus, these conclusions may not extend to the northern Lake.

SEDIMENT TRANSPORT

Sediment transport is a key physical process for the transport and distribution of metals and nutrients attached to sediment particles entering the Lake. The movement of sediments also impacts lake algal productivity and ecosystem dynamics because these sediments can control the availability of nutrients. Lastly, sediment particles in the water column also have an influence on in-lake water quality, for example, by increasing light attenuation and reducing light availability for algal productivity. Because there is no regular collection or analysis of sediment data in CDA Lake, the section below discusses the few special studies that have illuminated sediment transport processes.

Studies of In-Lake Sediments

There are several approaches to understanding sediment transport and morphological change in CDA Lake. These include investigating changes in lake bathymetry over time, analyzing surface and subsurface sediment samples, analyzing sediment core geochronology, and measuring suspended sediments in the water column over time and space. Each of these approaches can shed light on where sediments are moving within the Lake and whether they pass downstream of the Lake. Data on sediment grain size distribution, organic carbon content, and attached nutrients and metals are essential to understanding how suspended and deposited sediments influence the overlying water column chemistry, including algal productivity.

Bathymetry

Changes in lake bathymetry over time can be an indication of sediment erosion and deposition occurring in different areas of a lake. In order to interpret differences between bathymetric datasets, it is important to understand the spatial resolution and the accuracy of the data for each time period to assess the appropriateness of a comparison. Unfortunately, there appears to be limited bathymetric data over time for CDA Lake. Bathymetric surveys were conducted in 1994 by the USGS (Woods and Berenbrock, 1994), as shown in Figure 4-12, and in 2004 on behalf of Avista Corporation as part of its Federal Energy Regulatory Commission (FERC) relicensing process (CDA Tribe, 2004), as shown in Figure 4-13. The 1994 bathymetric survey consisted of only 561 data

points used to create the map, which limits the ability to interpret the accuracy of the depth contours. The 2004 bathymetric survey consists of thousands of data points based on the geographic information system (GIS) data layer (CDA Tribe, 2004). As a result of this disparity, it is not possible to do a quantitative comparison between the data sets and identify erosion and depositional areas. Qualitatively, there appears to be little difference between the maps, and, given a rough sedimentation rate of 4 cm/decade (Horowitz et al., 1993), it is not surprising that there is little difference over the ten-year time period between surveys. In order to assess if some of the Lake bottom has filled in or eroded over time, high-resolution bathymetric surveys are needed periodically.

The bathymetric maps are useful for a few qualitative interpretations. The general flow direction in the Lake is from south to north from the St. Joe and CDA Rivers to the outlet at Post Falls Dam. Both figures show that where constrictions in the lake bathymetry occur, the lake deepens. The deepest parts of the Lake are along the center, with shallower areas along the sides and in the various embayments. These indicators suggest that the predominance of the flow is along the middle of the Lake, which is expected, but vertical velocity profiles are unavailable. The sides of the Lake and the embayments would be expected to have slower velocities and therefore be more depositional to sediments than the main channel. There are seasonal variations in flow (see Chapter 3), and in suspended sediments entering the Lake. These seasonal variations will impact where sediments entering the Lake will be deposited, whether sediments will travel farther into the Lake, or whether they will pass downstream. In order to better understand sediment transport in the Lake throughout the year, suspended sediment and velocity profiles would need to be collected several times over the year.

Sediment Sampling

Horowitz et al. (1993) collected surface sediment samples in CDA Lake (see open dots in Figure 4-14), the CDA River, and the St. Joe River in 1989 and additional river bank samples in 1991 from the South Fork of the CDA River and the mainstem CDA River. Surface sediments in the Lake from south of Rockford Bay to the mouth of the St. Joe River contained an extremely cohesive, light gray silty clay grain size layer 1.5–3 cm below the surface—representing the ash from the Mount St. Helens eruption in 1981, which implies a recent sedimentation rate of 0.3–0.5 cm/yr. St. Joe River sediments were observed to be substantially coarser than the Lake surface sediments, with fine to medium sand with some silt indicating a clear distinction between sources. The CDA River sediments were observed to be coarser-grained than in the river delta and in the Lake itself but were characterized as finer-grained than material along the banks of the river.

Horowitz et al. (1993) measured the Lake sediment samples for silver (Ag), arsenic (As), copper (Cu), cadmium (Cd), mercury (Hg), lead (Pb), antimony (Sb), and zinc (Zn) and found the chemical distribution patterns in the Lake surficial sediments to be consistent with the CDA River sediments as a major source. Highest concentrations were found adjacent to the CDA River delta, which is consistent with the reduction in water velocity as the river enters the Lake, reducing sediment transport capacity and the sediments dropping out of the water column. However, high trace element concentrations in sediments from the CDA River delta south to Conkling Point and well within Windy, Rockford, Mica, and Wolf Lodge Bays were also observed. Horowitz et al. (1993) suggested there may be secondary source(s) of some of the high trace elements or physical factors other than the predominant flow direction (south to north), such as wind generated wave/currents remobilizing fine grained trace element-rich sediments. In reviewing the data presented in Horowitz et al. (1993, Table 3), the sediment size fraction < 2 μm constituted the higher trace element concentrations farther from the CDA River delta, while the sediment size fraction of 8–16 μm constituted the highest trace element concentrations near the CDA River delta. These observations support the notion that fine-grained materials from the CDA River enter the Lake and get transported throughout the Lake due to high inflows passing through the Lake, wind-driven wave action remobilizing sediments, remobilization from local currents in the Lake, or some combination of the above.

In a complementary study, Horowitz et al. (1995) analyzed subsurface sediments from 12 gravity cores collected in CDA Lake in June 1990 (see black dots in Figure 4-14). The majority of these cores were observed to have a heavily banded upper section and a lower, more homogenous section. The banded sections varied in thickness and number, with the authors stating this was an indication of sediments settling over time and being remobilized. Sediment cores 9 (Windy Bay) and 10 (Wolfe Lodge Bay) had lower metals concentrations in the banded sections

TABLE 4-1 Sedimentation Rate Estimates for Core 123

SOURCE: Horowitz et al. (1995).

than in the other cores. This may be due to the dominance of sediment deposition from small local streams and limited transport of sediments from the CDA River to these areas. Lower metals concentrations were observed in the banded section of the core at Site 8 (south of the CDA River), perhaps because St. Joe River flows prevented sediments from the CDA River from depositing (Horowitz et al., 1995).

Trace element data from the sediment cores were used to estimate that the total mass of metal-enriched sediment in CDA Lake is 75 million metric tons of sediments covering 85 percent of the Lake bed, which would indicate significant sediment transport through the Lake over time.

Sediment core 123 was used to conduct a geochronology analysis using cesium-137 activity. Considering a constant rate of deposition between dated points and no compaction, the sedimentation rates in Table 4-1 were estimated. The authors assumed the remainder of the sediment core (58 cm) had a sedimentation rate of 1.35 cm/yr, which means the undated core section represents 43 years and puts the start of the metal-enriched sedimentation in 1911, which is just after the 1880–1890 time period assigned to the onset of mining (Bender, 1991). Bookstrom et al. (2013) further interpreted sediment core 123 from Horowitz et al. (1995) and estimated that the period before 1895 had a sedimentation rate of about 0.1 cm/yr.

Once sediments enter CDA Lake, they tend to spread out based on local flow velocities and wind-driven wave action. This is demonstrated through the in-lake sediment sampling that indicates that trace element concentrations are highest near the mouth of the CDA River and decrease moving away from this location. However, high concentrations can still be found at some of the more secluded bays along the Lake.

Bookstrom et al.’s (2013) analysis of the sediment coring indicates that the net sedimentation rate has gone from 0.1 cm/yr at the turn of the 20th century to 2.1 cm/yr in the 1980s to 1990s, with higher sedimentation rates in the Lake bays than along the Lake axis. The reduced sedimentation along the Lake axis may be due to the Lake bathymetry and general direction of the Lake flow from south to north. The bathymetry surveys from 1994 and 2004 show the Lake has several constrictions along the south-to-north axis along the predominant flow path, and at these constrictions the Lake is deeper, likely maintained by this predominant flow path.

To confirm this conceptual site model of sediment transport in CDA Lake, additional data would be needed on suspended solids concentrations and grain-size distributions in the Lake, from the tributaries into the Lake, and in the flows out of the Lake. In addition, velocity data with depth in several places throughout the Lake—like along the main flow through the Lake, south of the CDA River, and in the bays—are key. Because sediments moving through the Lake carry organic carbon, nutrients, and metals, predicting future changes in Lake water quality will require better understanding of sediment quantities, physical and chemical characteristics of sediments, and how sediments interact biogeochemically across the Lake.

RESERVOIR MODELING OF COEUR D’ALENE LAKE

In lieu of data analysis, model simulations can provide insight into the processes at work in CDA Lake. Two models are discussed below, both of which consider in-lake processes that go beyond hydrodynamics and sediment transport. However, these models are discussed here because the data used to calibrate and validate the models were the C5 profile data discussed above. As argued below, both models would require substantial development before they could be used to simulate and accurately predict in-lake processes.

Aquatic Ecosystem Model 3D

The Aquatic Ecosystem Model (AEM3D) is a coupled hydrodynamics and biogeochemical model used for simulating lakes, reservoirs, estuaries and coastal oceans and is sold and supported by the company HydroNumerics.² AEM3D is based on the Estuary, Lake and Coastal Ocean Model–Computational Aquatic Ecosystem DYnamics Model (ELCOM-CAEDYM) developed by the Centre for Water Research in Australia. ELCOM-CAEDYM was used to model CDA Lake in 2007, using data collected in the dedicated field experiment conducted in 2005 and described earlier, along with longer-term seasonal data. In what follows, the model names ELCOM and CAEDYM are used instead of AEM3D, as those were the model names at the time the modeling was performed. AEM3D is presumed to have the same functionality.

ELCOM-CAEDYM can be used to simulate lake processes over timescales of days to years and at spatial scales of meters to kilometers. It includes a library of algorithms that represent many biogeochemical processes influencing water quality under the simulated physical conditions, including primary production; nutrient, carbon, and metal cycling; oxygen dynamics; and the transport and deposition of suspended solids.

The Hydrodynamic Model (ELCOM)

The hydrodynamic model ELCOM (see, e.g., Hodges et al., 2000) solves the unsteady Reynolds-averaged Navier-Stokes equations. ELCOM was run with data from May 20, 2005, to June 9, 2005, with a time-step of 300 seconds. The Coeur d’Alene bathymetry used for the initial simulations consists of a horizontal orthogonal grid with a constant horizontal grid size of 250 m. The vertical grid size is a uniform 1 m for the top 30 m of the water column and increases below this to 4 m at the bottom. The river-forcing data used were water temperature and flow rate data from the USGS gaging stations 12413860 (Coeur d’Alene River near Harrison), 12415140 (St. Joe River near Chatcolet), and 12419000 (Spokane River at Post Falls Dam).

The hydrodynamic model results were compared to high-spatial-resolution thermistor chain data and vertical profiles (described previously) that were acquired over a time period generally coinciding with the modeled period. The simulated temperatures captured some, but not all, of the measured features; no quantitative estimate of the quality of the match was provided. The 250-m model grid was unable to resolve the complex internal wave structure seen in the measured temperature field. Thus, it is uncertain the extent to which the model results can be relied upon to inform about the presence and absence of specific hydrodynamic processes (such as sediment resuspension, internal wave shear, and vertical diffusion). Although the model predicted currents and used these to produce estimates of the transport of modeled tracers, no water velocity data were acquired, and hence the fidelity of these modeled currents is unknown. The fact that the model was run for a relatively short time period (less than three weeks) and during spring also questions its ability to inform about processes occurring under the more strongly stratified and less inflow-dominated summer and fall periods.

A longer-term simulation (February 2003–December 2004) was conducted using straightened bathymetry with a 1,000 m × 250 m grid. Although the simulation results generally captured the onset and breakup of stratification and the overall Lake water balance, there was again no evidence that specific hydrodynamic processes were accurately represented.

The Biogeochemical Model (CAEDYM)

The objective for the application of CAEDYM to CDA Lake was to explore how the Lake may respond to altered loading of zinc and nutrients from the dominant inflows. The CAEDYM simulation includes descriptions of nutrient cycling and phytoplankton dynamics and operates at the same spatial and temporal scale as ELCOM. In addition to the base CAEDYM functionality, extensions were made to allow for more detailed simulation of heavy metals (in particular, zinc) within CDA Lake. The developments included the aqueous speciation of metals in response to changing geochemical conditions, inclusion of benthic fluxes of heavy metals from the sediment to the water column, uptake of metals into the algal biomass (and the associated toxicity effects), recycling of metals incorporated within the algal biomass during algal senescence, and sedimentation of heavy metals accumulated within algal or detrital material.

___________________

²https://www.hydronumerics.com.au/#about

The model was configured to include four phytoplankton groups (cyanophytes, chlorophytes, cryptophytes, and diatoms), based on parameters from a literature review. In addition to the phytoplankton, the simulation included organic matter, nutrients, and relevant inorganic ions. Fish, zooplankton or bacteria were not included in the model setup.

A new aqueous geochemistry sub-model was incorporated to allow for simulation of chemical reactions based on a user-defined set of elements (or components). The module is based on the equilibrium chemistry of aqueous solutions and their interactions with minerals and gases. The sub-model dynamically linked with the model’s biological cycles such that any biological activity (e.g., algal nutrient uptake and photosynthesis) could dynamically affect the aqueous speciation. As part of the solution process, pH, ionic strength, and other variables are also calculated.

Model Application to Coeur d’Alene Lake

The ELCOM-CAEDYM model was applied to CDA Lake for 2004, as the best data coverage existed at that time. The results, as reported by Hipsey et al. (2007), did not demonstrate any visual similarity between modeled and measured water quality variables such as dissolved oxygen. No statistical measures of model accuracy were provided, but since field data were available for only six days over the 11-month simulation period, it is doubtful that any such measures would have been meaningful. The predicted values for several of the more conservative cations and anions, such as calcium and chloride, gave some level of agreement with observations, as shown in Figure 4-15. However, predictions of more important variables, such as zinc, showed a questionable match to the data, with no predictive ability for upstream sites (orange and red symbols in Figure 4-15). Although correlations of zinc, sulfate, chloride, and calcium based on all data appeared to provide reasonable values of R² (in the range of 0.4–0.6), the zinc and sulfate values, in particular, appeared to show little correlation between measured and modeled concentrations when looked at site by site.

ELCOM-CAEDYM is a potentially useful and powerful tool for both understanding many of the complex hydrodynamic and biogeochemical processes that are taking place within CDA Lake and for testing and informing management actions. However, based on the modeling studies that were performed in 2005–2007, the calibration and validation of both the hydrodynamic model (ELCOM) and the biogeochemical model (CAEDYM) are incomplete. This is in part due to the fact that there were insufficient data of the right types and for a long enough period of time. For the hydrodynamic model, where the results were able to demonstrate the ability to capture some of the complexity in this system, longer sets of time series data for temperature at three to four locations along the length of the Lake are needed. The thermistors should be distributed between the surface and the bottom at each site. A minimum of two years of data would be required to allow a full year of calibration to occur, with a separate year for validation. The collection of full water column velocity profiles would also be important to obtain for different seasons to compare with model simulations. Important processes such as sediment resuspension are largely dependent on bottom current velocities.

For the biogeochemical model, a well calibrated and validated model could play an important part in better understanding some of the knowledge gaps about the CDA system. Some key parameters for which the data could be used in conjunction with temperature data are dissolved oxygen, conductivity, turbidity and pH, all of which can be collected with fast-response probes. Dissolved oxygen is possibly the most important driver for water quality, and having two years of continuous data would assist not only the modeling, but also the fundamental understanding of the biogeochemical processes themselves. The nutrient, phytoplankton, and metals data that need to be collected by direct water sampling would benefit from monthly sampling (over 12 months) along the entire Lake. Combined with the continuous temperature and dissolved oxygen data, this would go a long way to providing precisely what is needed to turn the models into useful tools for both management and science. As described in Chapter 10, some of the largest uncertainties have to do with future climate change impacts, and a reliance on models needs to be a part of addressing those uncertainties.

CE-QUAL-W2 Reservoir Model

Other models have been used for CDA Lake, including the laterally averaged CE-QUAL W2 model that was used by Avista Corporation, which owns and operates the Spokane River Hydroelectric Project. This project consists of the Post Falls, Upper Falls, Monroe Street, Nine Mile, and Long Lake Hydroelectric facilities. Avista used CE-QUAL-W2 as part of the process to renew its FERC license in 2005. Avista’s goal was to develop

CE-QUAL-W2 for CDA Lake to evaluate potential changes in water quality that could result from its lake-level operations (Golder Associates, Inc., 2005). Participants in the relicensing process, including the Washington Department of Ecology and the Idaho Department of Environmental Quality, expressed the need for a better understanding of the effects of project operations on water temperature, dissolved oxygen, pH, aquatic plants and nutrients. These agencies needed sufficient information to establish reasonable assurance that the project would comply with current provisions of the Clean Water Act.

The CE-QUAL-W2 model framework (Cole and Wells, 2016) is a two-dimensional, longitudinal/vertical, hydrodynamic and water quality model. Because the model assumes lateral homogeneity (i.e., it is laterally averaged), it is

best suited for relatively long and narrow waterbodies exhibiting longitudinal and vertical water quality gradients. The model has been applied to rivers, lakes, reservoirs, estuaries, and combinations thereof, including entire river basins with multiple reservoirs and river segments. The model framework includes hydrodynamics and aquatic biology and chemistry.

Although CE-QUAL-W2 is a well utilized and robust model framework for understanding water quality in surface water systems, its application in CDA Lake suffers from two key limitations that influence the model’s utility for considering the long-term water quality impacts. The first limitation is that the model framework is laterally averaged, which limits the ability of the model to consider 3-D effects in the Lake; medium to large lakes, such as CDA Lake, are inherently 3-D. The second limitation is that the CE-QUAL-W2 model framework does not have a full sediment transport model and it does not track the sediment size fractions that are deposited in a surface water system. This limits the model’s ability to accurately simulate sediment transport in the Lake and the nutrients, metals, and other chemicals associated with the sediment particles. Hence, the 3-D model of ELCOM-CAEDYM provides a more robust modeling framework for capturing hydrodynamics, sediment transport, and aquatic biogeochemistry in CDA Lake.

CONCLUSIONS AND RECOMMENDATIONS

This chapter analyzed the limited hydrodynamic data from CDA Lake, along with special studies on hydrodynamics and sediment transport in CDA Lake. The following detailed conclusions and recommendations are made.

1. The limited data available on specific conductivity in the river inflows and CDA Lake showed that water from the St. Joe River entered CDA Lake as an interflow, meaning that, because of its density relative to the stratification of the Lake, it intrudes below the thermocline but above the Lake bottom. The CDA River inflow, on the other hand, was classified as an overflow, intruding into the epilimnion above the thermocline. Hence, the two major river inflows are not likely to play a major role in resuspending sediment within CDA Lake. Sediment resuspension could, however, occur in the littoral zones of the lake due to other factors. The identification of those areas most subject to sediment resuspension, together with the relevant water quality impacts, can only be ascertained with a nearshore monitoring program, something that currently does not exist (see Chapter 8).
2. The committee’s analysis of conductivity data found that a St. Joe River inflow of 1,000 cfs is a threshold below which thermal stratification commences and internal lake processes will dominate over riverine influences. At river inflows above 1,000 cfs, which occur during winter and early spring, river discharge controls water quality (dissolved oxygen and pH) in CDA Lake, and the Lake behaves like a run-of-the-river system with little opportunity for biogeochemical processes to become established. At St. Joe River inflows below 1,000 cfs, which generally occur in June or later, internal dynamics and thermal stratification become important for CDA Lake, especially at site C5. It is during this period that water column processes such as nutrient uptake, phytoplankton proliferation, and decomposition can happen.
3. More data are needed to characterize thermal stratification, a key variable that controls hypolimnetic dissolved oxygen concentration and pH and ultimately sediment–water interactions. Stratification impacts all internal processes, including chemical and ecological processes. High-resolution time series temperature data are needed, preferably from four or five thermistor chains installed from south to north along the Lake axis. Instruments should be positioned from top to bottom along each chain. Ideally, profile measurements of dissolved oxygen, turbidity, conductivity, and pH should be collected at those sites. If the 1,000 cfs threshold for St. Joe inflow is reached earlier because of climate change, stratification will start sooner and one will observe a longer reduction of dissolved oxygen and lower pH (see Chapter 10).
4. There is a lack of data and information on sediment transport in the Lake, including the sediment particle size distribution at various locations. These are key drivers for the metal concentrations in lake bed sediments and mixing of sediments from various sources. This information can elucidate where sediments from the CDA River end up in the Lake and when and how sediments get transported downstream of the Lake. New data

4. collection should include the following data in the water column: vertical velocity profiles, thermistor vertical arrays at several locations, dye studies during low-flow and high-flow events, and particle size distribution data along with total suspended solids data.
5. There is a critical lack of deterministic model usage for the Lake—for both heuristic and predictive purposes—despite having invested in the first stages of developing such a model (ELCOM-CAEDYM). Development of a powerful 3-D hydrodynamic and water quality model began more than 15 years ago, but little has been done to use the model to better understand key processes within the Lake, the evolution of changes within the Lake, and the likely trajectory of the Lake under future climate changes. All parties have a shared interest to properly calibrate and validate the model with the goals of integrating the various processes that are at play in a complex lake. An initial thrust would be to understand dissolved oxygen dynamics and mass balances within the Lake. Such 3-D models can illuminate what is unknown about the system and what additional data collection could reduce these uncertainties.

REFERENCES

This page intentionally left blank.