Scientific Investigations Report 2008–5093
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2008–5093
An understanding of flow types, stage, river discharge, hydraulic geometry, and streambed sediments of the Coeur d’Alene River is necessary before developing hydraulic and sediment-transport models.
The elevation of Coeur d’Alene Lake affects the water-surface elevation of the Coeur d’Alene River in the study reach. The lake outlet (Spokane River) is restricted by a narrow, elevated, bedrock gorge—the lake is in a naturally dammed river valley. Post Falls Dam (RM 101.7), about 9 mi downstream of the inlet (fig. 1), augments lake storage and raises lake levels especially during summer. The dam controls lake outflows and levels from mid-June to mid-November (summer and autumn) (Avista Corporation, 2005). Post Falls Dam provides hydroelectric power, flood control, and irrigation supply. The lake and (or) dam cause water to back up (increase water-surface elevation) into the Coeur d’Alene River. When flows are high on the Coeur d’Alene and St. Joe Rivers (main tributaries), the lake level rises because inflow exceeds outflow. As high flows recede, outflows may exceed inflows and the lake level falls.
Two types of flow occur in the Coeur d’Alene River in the study reach—backwater and free-flowing water. Backwater is water backed up or retarded in its course, as compared with its normal or natural condition of flow (free-flowing water), usually caused by a downstream lake or dam or by a channel constriction downstream. Backwater also causes the upstream flow depth to increase. Backwater conditions are prevalent in the Coeur d’Alene River from the mouth to Cataldo Mission. Hypothetical water-surface curves illustrate backwater and free-flowing water conditions (fig. 3). Hydraulic engineers refer to the backwater curve as an M1 curve and the free-flowing water curve (above critical depth) as an M2 curve. Woodward and Posey (1941), Chow (1959), and Henderson (1966) describe such curves. Elongating the upstream channel in figure 3, would cause the backwater and free-flowing water curves to converge to the normal-depth curve. At the point where the backwater curve (M1) approaches or converges to the normal-depth curve, the effects of backwater cease. The normal-depth curve is sometimes called the no-backwater curve (Davidian, 1984). Backwater conditions do not affect any water-surface curve at or below the normal-depth curve. Figure 3 also illustrates that high downstream water-surface elevations cause the convergence point with the normal-depth curve to be farther upstream. When downstream water-surface elevations are high, the influence of backwater on the reach moves upstream.
In natural channels like the Coeur d’Alene River, the curves are not as smooth as shown in figure 3. Flow depth changes from point to point along a channel in response to differences in channel shape, slope, and roughness. Figure 4 shows backwater, normal depth, and free-flowing water curves for the Coeur d’Alene River in the study reach for a model simulation of 25,000 ft3/s at seven water-surface elevations at the Harrison gaging station (12413860) (RM 134.636). In figure 4, two curves (2,120 and 2,122 ft) are below the normal-depth curve indicating the reach has free-flowing water for the specified discharge and downstream water-surface elevation. Five curves are above the normal-depth curve (backwater condition). The curve for the downstream water-surface elevation of 2,126 ft transitions from backwater conditions to free-flowing conditions near RM 141. Figure 4 inset shows three backwater curves transitioning from backwater conditions to free-flowing conditions between RM 160.5 and 162.5, a 2 mi reach. For example, the 2,132 ft curve (fig. 4 inset) transitions from backwater to free-flowing conditions at RM 161.3, about 0.5 mi downstream of Interstate Highway 90, and the 2,134 ft curve transitions about 700 ft upstream of Highway 90. These transitions occur about 30 mi upstream of the Coeur d’Alene River inlet to the lake.
Water-surface elevations at gaging stations were obtained by adding the river stage and datum at each site. For this study, the North American Vertical Datum of 1988 (NAVD 88) datum was used. River stages (or water-surface elevations) have been measured on the Coeur d’Alene River at Harrison since 1991, at Cataldo since 1920, at Enaville since 1940, and at Pinehurst since 1987. River stages at Rose Lake were measured from 1995 through 2000. When cross sections were surveyed during 2004, datums at gaging stations also were surveyed using a Global Positioning System (GPS) allowing conversion of all data to one common datum. This study used the vertical datum of NAVD 88. Table 1 shows datum elevations in National Geodetic Vertical Datum of 1929 (NGVD 29) and NAVD 88 at six gaging stations on the river and lake. Locations of gaging stations are shown in figures 1 and 2. Differences between NAVD 88 and NGVD 29 are not the same everywhere (table 1) because of the Earth’s irregular curvature. AVISTA Corporation and others use another vertical datum called “lake datum” in the Coeur d’Alene area. To obtain elevations in NAVD 88, add 0.80 ft to elevations in lake datum.
Since 1903, the Coeur d’Alene Lake gaging station (12415500) has measured stage, which reflects inflow and outflow changes for Coeur d’Alene Lake. Water-surface elevations in the lake show seasonal fluctuations—usually maximum in spring and early summer and minimum in the winter (fig. 5). From the time Post Falls Dam became operational in 1907 until 1941, Avista Corporation (2005) held the summer and autumn lake level at about 2,127.3 ft (2,126.5 ft in lake datum). From 1942 until 2005, the levels were 2,128.8 ft (2,128.0 ft in lake datum). The increase of 1.5 ft since 1942 probably was due to increasing the hydrogeneration at the dam (Hank Nelson, Avista Corporation, oral commun., 2005). The maximum water level for the period of record was 2,139.8 ft on December 25, 1933, and the minimum was 2,120.7 ft on September 12, 1905.
Water-surface elevations for gaging stations on Coeur d’Alene Lake and River through the backwater reach are similar (fig. 6) but are at different elevations, reflecting the variable water-surface elevation of the backwater curve (M1) throughout the reach. For example, when elevations in Coeur d’Alene Lake (12415500) increase, water-surface elevations also increase in the river at Harrison (12413860) and Rose Lake (12413810) gaging stations. Water-surface elevations at the Harrison gaging station are almost identical with elevations at the Coeur d’Alene Lake gaging station.
Stage-relations were developed between the Coeur d’Alene Lake gaging station and the Harrison and Rose Lake gaging stations (fig. 7) because stage-discharge relations cannot be developed at the Harrison and Rose Lake gaging stations due to backwater. The curves and equations were derived using simple linear regression methods. These equations relate the water-surface elevation in the lake to water-surface elevations at the respective gaging stations. Figure 7 shows the r value (correlation coefficient), i value (number of paired data points), and the period of record (POR) value for the paired data. The correlation coefficient is a measure of strength of the linear relation between two variables (Zar, 1998). An r value of 0 indicates that no linear association exists between the two variables, whereas, an r value of 1 or -1 indicates a strong linear association. The correlation coefficient of water-surface elevations for the lake and Harrison gaging stations was 0.99 (very strong correlation). The correlation of water-surface elevations for the lake and Rose Lake gaging stations was a little weaker (r = 0.92) but was still strongly correlated. The r value decreased as distance increased from Coeur d’Alene Lake, probably because of the influence of inflows from intervening drainages, influence of inflows and outflows between the lateral lakes and floodplain, and (or) effects of backwater in the river.
The simple linear regression curves in figures 7A and 7B fit the data reasonably well. The curve shown in figure 7B is not accurate when water-surface elevation in the lake is greater than 2,128 ft because the data trend in another direction from the regression curve. The relation in figure 7B can be improved by using several linear regressions or non-linear function(s) that describe the entire dataset more accurately.
The Coeur d’Alene River has no dam or flood control structures and is subject to seasonal and peak flows. Discharge data have been collected at three gaging stations in the study reach (fig. 8). Continuous data have been collected since 1940 at the Enaville gaging station, 1987 at Pinehurst, and 1912–13 and 1920 at Cataldo except from 1973 to 1986 when the gaging station was not in operation. Figure 8 shows annual and seasonal patterns of discharge. Discharge is highest in winter and spring due to rainstorms and (or) snowmelt runoff and lowest in late summer and autumn when flows rely on base flow conditions. Mean annual discharge at the Cataldo gaging station is about 2,500 ft3/s. At Enaville, mean annual discharge is about 1,900 ft3/s, and at Pinehurst, about 510 ft3/s (Brennan and others, 2006). At Cataldo, average spring runoff is about 15,000 ft3/s and summer discharges usually are less than 600 ft3/s (fig. 8). The highest annual mean flow (1974) was about 3,300 ft3/s and the lowest (1977) was about 600 ft3/s (Brennan and others, 2006). For the greatest flood of record (January 16, 1974), mean daily discharge and instantaneous peak flow were 50,000 ft3/s and 79,000 ft3/s, respectively. For this event, indirect computations were used to estimate the mean daily discharge and instantaneous peak flow because the gaging station was not in operation. The February 9, 1996, flood produced a mean and instantaneous peak flow of 56,000 ft3/s and 70,000 ft3/s, respectively (Brennan and others, 2006; updated from Beckwith and Berenbrock, 1996).
Borden and others (2004) noted the volume of discharge passing gaging stations on the NF (Enaville) and SF (Pinehurst) does not sum to discharge at Cataldo (QEnaville+ QPinehurst ≠ QCataldo). Usually, the discharge sum from the Enaville and Pinehurst gaging stations was less than at the Cataldo gaging station (QEnaville+ QPinehurst < QCataldo). Borden and others (2004) attributed these differences to additional river discharges from intervening drainages and to streamflow gains and losses from the ground-water system.
Peak flow (flood) estimates for selected recurrence intervals at gaging stations on unregulated and undiverted streams in Idaho were developed by Berenbrock (2002) using peak flow data through 1997. That analysis was statistically based and included the Enaville, Pinehurst, and Cataldo gaging stations. These flood estimates are used for various purposes, such as design of bridges, culverts, and flood-control structures, and the management and regulation of flood plains. For these purposes, peak flow recurrence intervals generally are 50 years or greater. However, there is increasing interest in peak flows (floods) with more frequent recurrence intervals especially for peak flows that mobilize streambed sediments. The peak flow that is probably responsible for creating and (or) maintaining the characteristic size and shape of the channel (channel-forming flow) has been designated as bankfull flow (Leopold and others, 1964; and Knighton, 1998). Without human influences to streams, bankfull flows generally range between the 1.5- and 2.5-year peak flow (flood) (Leopold, 1994). For simplicity, bankfull flow for this study was defined as the 2-year peak flow although actual flows needed to overtop the human affected banks and (or) levees is probably higher.
The 100-year and 2-year peak flow estimates from Berenbrock (2002) were updated for the Enaville, Pinehurst, and Cataldo gaging stations to include an additional 10 years of peak flow data since 1997. The resulting estimates for the 100-year peak flows at Enaville, Pinehurst, and Cataldo gaging stations are 56,200, 14,300, and 69,000 ft3/s, respectively, and updated estimates of the 2-year peak flow were 15,000, 3,510, and 18,900, respectively. These values were not significantly different (less than 5 percent difference) from Berenbrock (2002) except for the 100-year peak flow at Pinehurst. One reason for the large difference at the Pinehurst gaging station is because 20 peak flow values (1988–2007) were used for this study as compared to 10 values (1988–97) used by Berenbrock (2002). Peak-flow data for the 10 years of record may have been collected during an unusually dry, wet, or otherwise unrepresentative period, and the data may not represent the full range of potential floods at the site; whereas, 20 years of record may be more representative.
The 1D sediment-transport model (HEC-6) requires accurate representation of the cross-section geometry. A series of cross sections measured at variably-spaced distances along lines oriented perpendicular to the flow direction defined cross-section geometries. Cross sections were surveyed to a common datum. Horizontal control was based on North American Datum of 1983 (NAD 83), Idaho Transverse Mercator Coordinates. Vertical control was based on NAVD 88.
In 2004, USGS personnel surveyed about 60 cross sections on the Coeur d’Alene River between RM 159.8 at Mission Flats to the Enaville (near RM 168.7) and Pinehurst (near RM 1.6) gaging stations. Appendix A shows the locations of cross sections. A GPS and echo sounder were used to obtain bathymetric data, and a GPS was used to obtain bank data along cross sections. The GPS and echo-sounder data were merged into one dataset and edited using HYPACK MAX software by Coastal Oceanographics. Cross sections were located to best represent the hydraulic characteristics of the river. Cross-section spacing ranged from about 800 ft to as much as 3,000 ft. To develop stream channel cross sections, the steps used by Barton and others (2004) were followed; however, a genetic computer program (Berenbrock, 2006) was used to reduce the number of echo (depth) soundings per cross section from hundreds or several thousands to 80 or fewer. Berenbrock (2006) showed that cross sections produced by the genetic algorithm were representative of the original section and processing time was significantly shorter than using standard procedures. The bathymetry for six cross sections on the SF Coeur d’Alene River was surveyed with a GPS because the river could be waded at these sections during the survey. Also, six cross sections in the braided reach from previous studies were used: four sections from the FourPt model (Woods and Beckwith, 1997) and two sections from the MIKE11 model (Borden and others, 2004). An indirect measurement from the flood of February 9, 1996, was used to develop the upper most cross section on the SF (cross section 1.580).
Several cross sections in the braided reach from the FourPt and MIKE11 models (Woods and Beckwith, 1997; Borden and others, 2004) were compared with surveyed sections from this study to ensure that they were compatible. These comparisons showed that cross-section geometries have changed very little. Figure 9 compares cross sections from the FourPt and MIKE11 models, respectively, with surveyed sections. Because of the similarity, it was acceptable to use cross sections from the previous FourPt and MIKE11 models from the braided reach in this study.
Cross sections in the meander reach (RM 159.8 near Mission Flats to RM 134.6 at the Harrison gaging station) were developed from gridded and LIDAR data. These cross sections (more than 160) are shown on a set of maps in appendix A. The gridded data from Avista Corporation are a composite of bathymetric data and digitized shoreline elevations from aerial photographs obtained at lake elevations of 2,126 and 2,128 ft in lake datum. The bathymetry was surveyed in 2004 using a dual, side-scan sonar. Control was set to the lake datum. LIDAR of the river and floodplain from Highway 95 to the lake were taken on August 31, 2004. Bathymetric data from Avista Corporation were used for the streambed part of the cross section. However, when the cross section extended above an elevation of 2,124 ft on the river banks, LIDAR data were used to avoid inaccuracies in digitizing. The floodplain part of the cross sections was always determined using LIDAR data.
The following steps were used to develop cross sections in the meander reach (see appendix A for locations of cross sections). First, gridded data greater than 2,124 ft were eliminated. Second, the gridded and LIDAR data for each cross section were merged. Third, the data were ordered from left bank (viewed as looking downstream) to right bank, with the left-most data point designated as 0 ft. Distance for all other data points on the cross section extended from the left-most point. Again, cross-section data generally were reduced to 80 or fewer points per cross section using a genetic computer program (Berenbrock, 2006) designed for this purpose.
Each cross section was given a station number in river miles corresponding to its location on the river. River miles referenced in this report were derived from designations by the USGS National Mapping Division on 7.5-minute (1:24,000–scale) maps and by the Columbia Basin Inter-Agency Committee (1964). River miles generally are measured in 1-mi increments along the channel centerline in an upstream direction, beginning with 0 at the mouth of the Spokane River at the confluence with the Columbia River at Franklin D. Roosevelt Lake (fig. 1). Since the publication of the referenced documents, the channel length in several places along the Coeur d’Alene River has changed because of bank erosion, deposition, or channel migration. As a result, the actual distances between the river mile designators shown on the maps are not always exactly 1.0 mi. However, because many people are familiar with them and use them for various purposes, continued use is appropriate.
To use these river mile designators for identifying cross sections, the actual distance between river miles was measured from the National Agriculture Imagery Program (NAIP) digital orthophotos taken in 2004. Distance between the nearest downstream river mile designator to the cross section was measured along the channel’s centerline of the digital orthophoto. The ratio of cross-section distance to distance between river miles was calculated and used as the extension in the cross-section name. For example, cross section RM 156.667 is two-thirds of the way between RM 156 and 157.
Channel cross sections differ between the braided and meandering reaches. Cross sections in the braided reach have widths greater than 500 ft, depths usually less than 15 ft, and average depths less than 10 ft. At low flows, exposed islands and (or) bars at these cross sections divide the river into separate channels. High flows often submerge many of the islands and (or) bars. However, a few permanent islands are in this reach near RM 162.5 and RM 167. From RM 165.7 to 166.8 near Kingston, the channel splits into two branches with the main channel located in the left branch. Channel slope in the braided reach is about 6.5 × 10-4 ft/ft. Cross sections in the meander reach are about 300- to 400-ft wide with depths greater than 20 ft. Depths greater than 40 ft occur in and around channel bends. Channel slope in the meander reach is about 6.5 × 10-5 ft/ft, 10 times less than the slope in the braided reach.
Information on particle-size distribution came from streambed material samples collected by previous investigators (URS Greiner, Inc., and CH2M-Hill, 2001; Borden and others, 2004) and this study. In 1997, four sites were sampled to characterize the streambed sediments in the meander reach of the Coeur d’Alene River (URS Greiner, Inc., and CH2M-Hill, 2001). Streambed sediment samples were collected near the left bank (LB), left of middle of channel (LMID), right of middle of channel (RMID), and near the right bank (RB). These samples were collected using a drill core and sent to a laboratory for particle-size analysis. The d50 of these samples shown in table 2 ranged from 0.016 mm at RM 132.1 near Harrison to 0.34 mm at RM 159.6 near Mission Flats. They are classified as medium silt and medium sand, respectively. The d50 decreased downstream. These samples showed no significant variation in d50 across the channel.
In 2001, the University of Idaho (Borden and others, 2004) sampled five sites in the braided reach. Wolman (1954) described the sampling method used to determine the size distribution of sediments along the surface of the streambed. Samples included only particles larger than 4 mm (Borden and others, 2004) from the armored surface layer, the topmost layer. A ½ phi (φ) gravelometer, that measures the intermediate axis of a particle, was used to sample the armored-surface layer. Armoring is not present in sandbed dominated reaches, (for example, the meandering reach). Sediments in an armored surface layer commonly are larger than the underlying channel fill. These sediments tend to armor or protect the streambed from erosion under low- to moderate-flow conditions. The d50 of bed material below the armored layer ranged from 18.8 to 53.7 mm (table 2), and d50 decreased downstream. These samples were classified as coarse gravel to very coarse gravel (table 3). Particle-count analysis from Borden and others (2004) indicated the largest particle on the surface layer was 128 mm—the approximate thickness of the armored-surface layer. A particle of this size is considered coarse cobble (table 3). The smallest d50 was in the SF probably because of the additional sediment contribution or loading of fine sediments from mine tailings.
In 2005, four additional samples were collected to better define the streambed in the Dudley reach for use in a 1D sediment-transport model (HEC-6) and a 2D hydraulic and bed-shear stress model. These samples were collected using a ponar dredge sampler near the center of the channel. Particle-size analysis was conducted on these samples using the equipment and facilities at the Department of Engineering Laboratory, Boise State University. The methods and procedures used for the particle-size analyses are outlined in the American Society for Testing and Materials’ Manual on Test Sieving Methods (1985), the USGS Vancouver Sediment Laboratory’s Quality Control and Quality Assurance Plan (Daniel J. Gooding, U.S. Geological Survey, written commun., 2000), and the American Society of Civil Engineers’ Manual on Sedimentation (1975). These samples were analyzed using the dry-sieve method described in the references above to determine the percent-finer-than and the cumulative percent-finer-than fractions, and the particle-size characteristics.
Table 2 shows the d50 and particle-size and classification of the four Dudley reach samples. The cumulative percent-finer-than fractions and other particle-size characteristics are provided in appendix B. The d50 ranged from 0.31 to 0.59 mm (table 2) and was classified as medium to coarse sand, respectively. The d50 decreased downstream. In sample D1 (cross section 157.067), three large particles with diameters (measure of intermediate axis) larger than 5 mm (fine gravel) were interspersed in a matrix of moderately-sorted sand (appendix B). This sample also included particles ranging from 2 to 4 mm (very fine gravels). Therefore, the d50 classification of this sample was medium sand. One particle with an intermediate axis of about 8 mm (medium gravel) was collected in sample D2 (at cross section 156.686. This sample (classified as coarse sand) primarily consisted of poorly sorted sands and gravels and included the widest particle-size distribution of the dredge samples. Samples D3 and D4 were classified as medium sand.
U.S. Department of the Interior | U.S. Geological Survey
URL: http://pubs.usgs.gov/sir/2008/5093
Page Contact Information: Publications Team
Page Last Modified: Thursday, 10-Jan-2013 18:49:13 EST
