USGS Home
Contact USGS
Search USGS

U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2008–5168

Back to Table of Contents

Methods

Limnological Study, Calendar Years 1991–92

The 1991–92 study was conducted during 2 calendar years, beginning in mid-January 1991 and ending in mid-December 1992. Seven pelagic stations were sampled every 3 weeks from May through October and every 4 to 6 weeks during the remaining months. Pelagic stations were located in the open water, deeper (10 m or more) areas of the lake. A full description of the sampling design and rationale for the 1991–92 study is provided in Woods and Beckwith (1997).

Full-depth profiles of water column temperature, specific conductance, pH, and dissolved oxygen concentration and percent saturation were measured. Secchi-disc transparency depth was measured and the euphotic zone depth was determined based on water column profiles of photosynthetically active radiation (PAR) intensity. Within the euphotic zone, a composite sample was derived from three point samples that were collected with a non-metallic Van Dorn bottle. Two additional point samples were collected at mid-depth and 1 m above the lakebed. These water samples were analyzed for concentrations of total phosphorus, dissolved orthophosphate, total organic plus ammonia nitrogen, dissolved ammonia, and dissolved nitrite plus nitrate. Chlorophyll-a was determined for the euphotic zone composite samples. On alternating sampling trips, euphotic zone and near-bottom samples were analyzed for total recoverable concentrations of arsenic, cadmium, copper, lead, mercury, and zinc. The water samples were analyzed at the USGS National Water Quality Laboratory (NWQL).

Comparisons of pelagic (open water) and littoral zone (near shore) water quality were made on two occasions and were based on 2-m-deep littoral zone samples and euphotic zone composite samples from the pelagic stations. During September 1991, 20 littoral and 6 pelagic stations were sampled concurrently for nutrients, chlorophyll-a, and dissolved oxygen. During August 1992, 15 littoral and 6 pelagic stations were sampled concurrently for those same variables plus total zinc. Littoral zone sampling on both occasions covered stations throughout the lake (Woods and Beckwith, 1997).

No records could be located for quality assurance/quality control samples collected during the 1991-92 study. Therefore, no assessment could be made to determine if equipment decontamination procedures were adequate to remove analytes of interest or to quantify variability introduced during sample processing and analysis.

Limnological Study, Water Years 2004–06

Samples were collected by USGS crews at five stations in the pelagic zone (fig. 1; also sampled during 1991–92), eight times each year during water years 2004–06. A water year is from October 1 through September 30. As in the 1991–92 limnological study, full-depth profiles of the water column temperature, specific conductance, pH, and dissolved oxygen concentration and percent saturation were measured. Secchi-disc transparency depth was measured, and the euphotic zone depth was determined with water column profiles of PAR intensity. Within the euphotic zone, a composite sample was collected using a pump sampler lowered through the euphotic zone. Additional point samples were collected at depths representing the upper hypolimnion and 1 m above the lakebed representing the lower hypolimnion. Water samples were analyzed at the USGS NWQL.

Coeur d’Alene Tribe crews collected samples at 12 stations in the littoral zone on 4 occasions in water year 2004 and at 18 stations on 4 occasions in water year 2005 and early in 2006. These sites were primarily near the heads of bays, in water about 4-m deep, and generally within 20 m of the lakeshore (fig. 1). Point samples were collected from a depth of 2 m, where temperature, specific conductance, pH, and dissolved oxygen concentration and percent saturation also were measured. Coeur d’Alene Tribe crews sampled the littoral stations during the same week that USGS crews sampled the pelagic stations. Littoral zone samples were shipped by USGS to the NWQL for analyses.

Comparisons of pelagic and littoral zone water quality were made on two occasions and were based on 2-m-deep littoral zone samples and euphotic zone composite samples from the pelagic stations. During August 2004, 12 littoral and 5 pelagic stations were sampled concurrently for nutrients, chlorophyll-a, and metals. During July 2005, 18 littoral and 5 pelagic stations were sampled concurrently for the same variables.

Quality assurance/quality control samples were collected during routine sampling trips. Field blank samples were submitted to the NWQL at a frequency of 4 percent to assess whether equipment decontamination procedures were adequate to remove analytes of interest. Three source solution blanks were submitted to assess possible contamination in blank water used in the field blank samples. Duplicate or replicate samples were submitted at a frequency of 5 percent to assess variability introduced during sample processing and analysis.

The 2004–06 study approach was designed to complement previous limnological studies of the lake that were done mostly by the USGS since 1987 (Woods, 1989; Horowitz and others, 1993, 1995a, 1995b; Woods and Berenbrock, 1994; Woods and Beckwith, 1997; Balistrieri, 1998; Kuwabara and others, 2000; Woods, 2004). The study approach also incorporated important new limnological information obtained from the RI/FS of the lake conducted by the USGS for USEPA during 1999 in support of the Bunker Hill Superfund Site OU3 ROD (URS Greiner, Inc., and CH2M Hill, Inc., 2001; CH2M Hill, Inc., and URS Corp., 2001).

The RI/FS (URS Greiner, Inc., and CH2M Hill, Inc., 2001) and a USGS evaluation (funded by the National Water Quality Assessment Program) of fate and transport processes in Coeur d’Alene and Pend Oreille Lakes in northern Idaho and Flathead Lake in northwestern Montana (Woods, 2004) reported that lake hydrodynamics and riverine inflow plume routing played important roles in the fate and transport of constituents and, by extension, water-quality conditions in the lakes. Therefore, limnological data collection in Coeur d’Alene Lake during water years 2004–06 was scheduled during important periods related to lake hydrodynamics and limnological conditions.

The approaches used for the two studies were similar in many respects and allowed for robust comparisons between most analytes. Two differences in approach merit further discussion: (1) compositing of euphotic zone samples, and (2) analytical methods for determination of chlorophyll-a concentrations.

Water-quality profiling instrumentation used in the 2004–06 study provided high resolution (less than 0.1 m) data that revealed substantial variations in constituent concentrations within the euphotic zone, especially during periods of thermal stratification. This variability was represented in the pumped-sample compositing used in the 2004–06 study. The 3-point sample compositing used in the 1991–92 study was less likely to have obtained a representative sample of the euphotic zone.

Concentrations of chlorophyll-a for the 1991–92 study were determined with High-Performance Liquid Chromatography (HPLC) using methods from Britton and Greeson (1989). Prior to the start of the 2004–06 study, however, the USGS NWQL replaced the HPLC method with a chromatographic-fluorometric (C-F) method, presented in U.S. Geological Survey National Water Quality Laboratory (2000) and Britton and Greeson (1989). Reasons for the change in analytical methods were provided by U.S. Geological Survey, National Water Quality Laboratory, written commun., 2005. This change presented potential difficulties in comparing concentrations of chlorophyll‑a between the two studies. The approach presented in this report to quantify differences between the two methods relied on the results of a NWQL comparability study where results of the two methods could be correlated through a regression equation (R. Brenton and P. Soliven, U.S. Geological Survey, written commun., October 1995). The regression equation reported in the NWQL study was derived from 39 paired samples and had the following form, with a coefficient of determination of 0.771:

(C–F) = 0.45801 + 0.92047 (HPLC),      (1)

where:

Equation 1 indicated that C-F values of 1.4, 5.1, and 18, respectively, were associated with HPLC values of 1.0, 5.0, and 19.

In conjunction with the work described in this report, a lake process simulation model called ELCOM-CAEDYM was developed for Coeur d’Alene Lake by the Centre for Water Research of the University of Western Australia under a cooperative agreement with the USGS and IDEQ. The goal of the model development was to provide agencies involved with the lake management decision process with a sophisticated tool to simulate lake response to a wide range of mining associated contaminant remediation and nutrient management strategies likely to occur or be proposed under the Bunker Hill Superfund Site OU3 ROD and Lake Management Plan. A detailed description of the model that was adapted to Coeur d’Alene Lake is available at http://www.cwr.uwa.edu.au (Hipsey and others, 2006). The USGS limnological studies of 1991–92 and 2004–06 were primary data sources for the application of the model to Coeur d’Alene Lake.

Sampling Stations and Schedules

Pelagic zone sampling was conducted at five stations common to the 1991–92 and 2004–06 studies (fig. 1; table 1). Stations are identified by number as 1, 3, 4, 5, and 6. These stations have been sampled historically by USGS (Woods and Beckwith, 1997), IDEQ (G.F. Harvey, Idaho Department of Environmental Quality, written commun., 2000), and the USEPA during the RI/FS (URS Greiner, Inc., and CH2M Hill, Inc., 2001). Representative lake volumes were assigned to each pelagic station by digitizing a bathymetric map of the lake, developing area polygons by joining midpoints of lines drawn between stations, and calculating lake volume under each polygon footprint. According to this method, the total lake volume represented by each station is: station 1, 27 percent; station 3, 34 percent; station 4, 34 percent; station 5, 4 percent; and station 6, 1 percent. Although stations 5 and 6 together represent only 5 percent of the lake volume, they are important because they characterize the southern end of the lake, which is more productive than other parts of the lake. Water quality in this area is determined mostly by the St. Joe River inflow. Station 6 provides a reference condition for metals concentrations, which when detected, typically are low.

Limnological samples were collected at the five pelagic stations eight times each year during water years 2004–06. Timing of sample collection reflects temporal patterns associated with lake hydrodynamics, nutrients, lake productivity, and fate and transport of trace metals and nutrients (table 2). Dates of sample collection at pelagic station 4 (representative of the other four stations) and lake stage during water years 2004–06 are shown in figure 2.

Littoral zone sampling locations were not identical for the 1991–92 and 2004–06 studies. Two sets of stations were used for sampling during 2004–06. During water years 2004–06, 12 stations representative of bays throughout the lake (identified as “NS”) were sampled 4 times each year. In water year 2005, 6 additional stations were sampled in the southern end of the lake near seeps (identified as “S”) to assess trace-metal contamination adjacent to a former mining industry railroad upstream of the lake. Station locations and sample dates for the 18 littoral stations are listed in table 3 and shown in figure 1. Most littoral zone samples were collected at the 2-m depth in water columns that were about 4-m deep and within about 20 m of the shoreline.

Mass Balances of Nutrients and Trace Metals

The evaluation of nutrient and trace metal mass balance in this report was based on annual values derived from constituent load information generated for five USGS streamflow-gaging stations in Idaho: Coeur d’Alene River near Harrison (12413860); St. Joe River at St. Maries (12415075); St. Joe River near Chatcolet (12415140); Spokane River at Lake Outlet at Coeur d’Alene (12417598); and Spokane River near Post Falls (12419000) (fig. 1). Periodic hydrograph-based monitoring at the five gaging stations provided streamflow and concentration data required for load calculations. The 1991–92 study reported annually based hydrologic, nutrient, and trace metal budgets for Coeur d’Alene Lake (Woods and Beckwith, 1997). Three major differences between the 1991–92 and 2004–06 studies were: (1) additional constituents were measured in the 1991–92 study, and some common constituents had different method reporting limits in the 2004-06 study, (2) loads for the 1991–92 study published in Woods and Beckwith (1997) were calculated using the computer program FLUX (Walker, 1996) and loads for the 2004–06 study were calculated using the computer program LOADEST (Runkel and others, 2004), and (3) sampling locations changed on the St. Joe and Spokane Rivers between studies. For this report, only loads measured in common between the two studies were compared. The accuracy of the load budgets is affected by changes in method reporting limits after the NWQL updated analytical procedures in the late 1990s, and a robust direct comparison cannot be made for all constituents. The 1991–92 loads were recalculated using LOADEST so a proper comparison could be made with the 2004–06 loads. The LOADEST program uses a multiple linear regression approach to calculate loads based on available streamflow records, decimal time, seasonality functions, and other user-input explanatory variables (Runkel and others, 2004). The Adjusted Maximum Likelihood Estimator (AMLE) regression model in the LOADEST program was used to incorporate censored data with multiple reporting limits.

The gaging station on the St. Joe River was moved from St. Maries (12415075) to Chatcolet (12415140) between the 1991–92 and 2004–06 studies to capture total inflow to the lake from the St. Joe River. The station sampled during the 1991–92 study, 12415075, did not capture nutrient inputs from the St. Maries wastewater treatment plant because it was upstream of the plant outfall. Similarly, the gaging station on the Spokane River was moved from Post Falls (12419000) to Coeur d’Alene (12417598) between the 1991–92 and 2004–06 studies to quantify water-quality conditions at the primary outlet of the lake. Water-quality conditions at the station sampled in 1991–92, 12419000, were influenced by nutrients in effluent from the Coeur d’Alene, Hayden, and Post Falls municipal wastewater treatment plants. The gaging station sampled in 2004–06, 12417598, was upstream of these plant outfalls. The location of the gaging station on the Coeur d’Alene River, 12413860, remained constant between the study periods.

Nutrient load data for the Spokane and St. Joe Rivers in the 1991–92 study period were normalized to locations sampled during the 2004–06 study period for comparison of loads. Nutrient loads contributed by wastewater effluent were subtracted from the 1991–92 loads measured at station 12419000, Spokane River near Post Falls, to obtain a measure of nutrient loads at the lake outlet. Similarly, nutrient loads from wastewater effluent were added to the 1991–92 loads measured at station 12415075, St. Joe River at St. Maries. Trace metal loads were not normalized for sampling location because no data were available for computing additional inputs or losses, if any, between the sampling locations on the Spokane and St. Joe Rivers. Additional inputs or losses between sampling locations for trace metal loads are assumed to be negligible.

Total nitrogen and total phosphorus loads in effluent from the St. Maries wastewater treatment plant to the St. Joe River in 1991–92 were reported as 3,720 kg/yr and 1,400 kg/yr, respectively, in Woods and Beckwith (1997). Average monthly loads for total phosphorus in wastewater effluent to the Spokane River were calculated from records provided by the Coeur d’Alene and Hayden municipal wastewater treatment plants. Average monthly loads for total nitrogen effluent were calculated for the Coeur d’Alene plant. Nitrogen compounds were not monitored in effluent from the Hayden plant during 1991–92; therefore, records were not available for calculation of nitrogen loads. Nutrient inputs from the Hayden plant were small compared to nutrient inputs from the Coeur d’Alene plant, and wastewater effluent was discharged to a land application site, not directly to the Spokane River, for most months in 1991–92. Consistent monitoring records were not available for the Post Falls wastewater treatment plant in 1991–92, but the plant discharged to the Spokane River year-round during the 1991–92 study period (Mark Barkley, City of Post Falls, oral commun., July 2008). Therefore, nutrient loads from the Post Falls treatment plant were estimated by multiplying nutrient loads from Coeur d’Alene treatment plant effluent by a ratio between the populations of Coeur d’Alene and Post Falls in 1991–92, according to the following equation:

PF WWTP nutrient load = CdA WWTP nutrient load × (PF population/CdA population),      (2)

where

Annual loads of nutrients from ungaged surface-water inflows, including Plummer, Fighting, Carlin, and Wolf Lodge Creeks, were measured during the 1991–92 study and reported in Woods and Beckwith (1997). Ungaged surface-water inflows were not sampled during the 2004–06 study; however, nutrient concentrations in inflows from these ungaged inflows were assumed to be relatively constant between the 1991–92 studies because nutrient sources in these tributaries primarily are agricultural and natural sources, not municipal wastewater treatment plant effluent. Nutrient loads in ungaged surface-water inflows in the 2004–06 study were estimated for Plummer, Fighting, Carlin, and Wolf Lodge Creeks as well as for other ungaged surface water including small tributaries and runoff. Nutrient coefficients were calculated by relating the measured 1991 and 1992 nutrient load in a particular creek to the measured flow volume in the nearest gaged inflow, the Coeur d’Alene or St. Joe Rivers, for the same year. The 1991 and 1992 nutrient coefficients then were averaged. Inputs from “other” ungaged surface water inflows were related to the combined flow of the Coeur d’Alene and St. Joe Rivers as the exact locations of these sources are unknown. The equation for the average nutrient coefficient for ungaged surface-water inflows had the following form:

C_avg = ((L₉₁/Q_near91) + (L₉₂/Q_near92))/2,      (3)

where:

The average nutrient coefficient from the 1991–92 study was then used to estimate loads from ungaged surface-water inflows for each year during the 2004–06 study according to the following equation:

L = C_avg x Q_near,      (4)

where:

Annual nutrient loads from precipitation and other atmospheric sources in the 1991–92 study were estimated and published in Woods and Beckwith (1997) based on data presented by Stanford and others (1983), Hallock and Falter (1987) and National Atmospheric Deposition Program (1991). These values were assumed to be constant for the 2004–06 study. Annual nutrient loads from private and community wastewater-treatment systems in the nearshore area of the lake were estimated for the 1991–92 study and published in Woods and Beckwith (1997). These values also were assumed to be constant for the 2004–06 study because no recent data were available.

Annual nutrient loads leaving Coeur d’Alene Lake in ground-water outflow to the Rathdrum Prairie aquifer were computed by multiplying the estimated ground water outflow volume, in cubic hectometers, by the mean nutrient concentration, in micrograms per liter, in the lower hypolimnion of the northern end of Coeur d’Alene Lake. Ground water outflow volume was estimated and reported as 205 cubic hectometers per year for the 1991–92 study in Woods and Beckwith (1997) and was assumed to be constant for the 2004–06 study. The mean annual concentration of nutrients, in micrograms per liter, in the epilimnion of the northern end of the lake was multiplied by the annual change in lake volume, in cubic hectometers, to compute the annual nutrient load associated with this budget component. Change in lake volume was computed separately for each year in both studies.

Errors were calculated or estimated for each budget component. Standard errors for the load estimates computed for the St. Joe, Spokane, and Coeur d’Alene Rivers were provided in the LOADEST output as average daily errors for each month. These daily errors were multiplied by the number of days in each month and were then summed for each year in the 1991–92 and 2004–06 studies. Errors for nutrient loads from nearshore wastewater and precipitation for the 1991–92 study were reported in Woods and Beckwith (1997); because loads for these budget components were assumed to be constant for the 2004–06 study, errors also were assumed to be constant. Similarly, errors for nutrient loads from ungaged surface-water inflows, ground-water outflows, and lake storage change for the 1991–92 study were reported in Woods and Beckwith (1997). However, these loads were not assumed to be constant for the 2004–06 study. Because loads for these budget components in the 2004–06 study were estimated using relations with loads from the 1991–92 study, errors were assumed to be proportional. An average percent error relative to a load was calculated for the 1991–92 study and applied to the 2004–06 load estimates. For example, the average percent error (error divided by the associated load, in percent) in total nitrogen load in ground-water outflow for the 1991–92 study was 25 percent. Errors in total nitrogen loads in ground-water outflow for each year in the 2004–06 study were calculated by multiplying the annual load by the average percent error in the 1991–92 study (25 percent). Because recent data were unavailable to calculate loads for some budget components, reported errors for the 2004–06 study may not be accurate.

Overall error for each nutrient and trace metal budget was computed using the following equation (Brown, 1987):

OE = ((E₁)² + (E₂)² + … + (E_n)²)^0.5,      (5)

where:

Depth Profiles and Nutrient Sample Collection

Status and trends of lake productivity indices such as water-column transparency, nitrogen and phosphorus concentrations, chlorophyll-a concentrations, and dissolved-oxygen concentrations were evaluated using limnological data. Water-column circulation processes also were evaluated on the basis of spatial and temporal changes in nutrients, chlorophyll-a, and dissolved-oxygen concentrations.

Prior to collecting water samples at the pelagic stations, full-depth profiles of the water column were measured for temperature, pH, oxidation-reduction potential, specific conductance, dissolved oxygen concentration and percent saturation, light transmissivity (a surrogate for turbidity), and fluorescence (a surrogate for chlorophyll-a). The upper water column was profiled for PAR to define the euphotic zone depth. Secchi-disc transparency depth was also measured. The euphotic zone is the part of the water column in which PAR is equal to or greater than 1 percent of the PAR incident on the lake surface. Water column samples were collected eight times per year at the five pelagic stations. At the deeper stations (1, 3, and 4), samples were collected in the euphotic zone (composite), upper hypolimnion, mid-hypolimnion, and lower hypolimnion (1 m above lakebed). Station 5 was sampled in the euphotic zone and upper and lower hypolimnion, whereas station 6, the shallowest, was sampled in the euphotic zone and lower hypolimnion. The samples were analyzed for total concentrations of nitrogen and phosphorus and dissolved concentrations of nitrite plus nitrate, ammonia, phosphorus, and orthophosphate. The euphotic zone samples also were analyzed for concentrations of chlorophyll-a and pheophytin. Littoral zone samples were analyzed using the same methods as pelagic euphotic zone samples, except only one 2-m-deep sample was collected at each station.

Trace Metal Sample Collection

Limnological data were evaluated for the status and trends of trace metal concentrations in the lake. Water column circulation processes also were evaluated on the basis of spatial and temporal changes in trace metal concentrations.

Water column samples for trace metal analyses were collected in conjunction with pelagic zone nutrient samples during both study periods. The samples from the five pelagic stations were analyzed for total and dissolved concentrations of cadmium, lead, zinc, iron, and manganese, as well as hardness. The lower hypolimnion samples also were analyzed for dissolved arsenic. Littoral zone samples were analyzed for the same constituents as the pelagic euphotic zone samples, except only one 2-m-deep sample was collected at each station.

The exchange of constituents between a water column of a lake and lakebed sediments, often termed benthic flux, has been identified as an important water-quality issue for Coeur d’Alene Lake (Horowitz and others, 1993, 1995a, 1995b; Kuwabara and others, 2000; and Woods, 2004). To complement water column sampling and to obtain recent information related to the benthic flux of trace metals, samples near the sediment-water interface were collected at station 4 on each sampling trip during the 2004–06 study. To avoid disturbing the thin veneer of flocculent material at the sediment-water interface, samples were collected by gentle and slow insertion of a modified gravity coring device (with non-metallic core tube liner, core cutter, and core catcher) into the lakebed sediments. Upon retrieval, the water within 0.1 m of the lakebed (but above the flocculent layer) was gently removed from the core tube liner using a peristaltic pump. The sample was discarded if significant re-suspension of sediment was observed or suspected during retrieval. Analyses for trace metals and nutrients were the same for these samples as the samples from the water column. Analyses also included particulate and dissolved organic carbon (DOC).

Back to Table of Contents