USGS Home Contact USGS Search USGS

Scientific Investigations Report 2005–5030

In cooperation with the Menominee Indian Tribe of Wisconsin

Trends in Streamflow, Sedimentation, and Sediment Chemistry for the Wolf River, Menominee Indian Reservation, Wisconsin, 1850–1999

By Faith A. Fitzpatrick

The PDF for the report is 3,926 kb

Table of Contents

Abstract

Introduction

Methods

Streamflow Trends

Sedimentation Trends

Sediment-Chemistry Trends

Summary and Conclusions

Acknowledgments

References Cited

Figures

Figure 1. Location of valley transects, stream gages, and precipitat...

Figure 2. Photos of A, Balsam Row Dam in 1998, looking upstream at d...

Figure 3. Longitudinal profile of the Wolf River through the Menomin...

Figure 4. Boxplots of A, mean daily streamflow, and B, annual peak f...

Figure 5. Annual peak streamflow from the Wolf River near Shawano (t...

Figure 6. Annual peak flow and trend lines for the Wolf River near S...

Figure 7. Annual precipitation and trend lines for Green Bay, Wis., ...

Figure 8. Trace of the Wolf River, Wis., showing channel position in...

Figure 9. Detailed longitudinal profile of the Wolf River from the B...

Figure 10. Valley transect through the impounded reach of the Wolf R...

Figure 11. Valley transect through the impounded reach of the Wolf R...

Figure 12. Valley transect through the impounded reach of the Wolf R...

Figure 13. Data for ¹³⁷Cs, unsupported ²¹⁰Pb, and ...

Figure 14. Profiles for A, particle-size distribution, B, bulk densi...

Figure 15. Valley transect through the impounded reach of the Wolf R...

Figure 16. Valley transect through the sandy reach of the Wolf River...

Figure 17. Partial valley transect through the sandy reach of the Wo...

Figure 18. Partial valley transect through the sandy reach of the Wo...

Figure 19. Valley transect through the sandy reach of the Wolf River...

Figure 20. Valley transect through the rocky reach of the Wolf River...

Figure 21. Ice damage on trees in flood plain near transect T9, Apri...

Figure 22. Selected profiles of trace elements for core T2-1 from th...

Tables

Table 1. U.S. Geological Survey stream gages used in the analyses of...

Table 2. Description of valley transects and cores collected along t...

Table 3. Summary of core locations, water depths, penetration depths...

Table 4. Selected trend-test results for streamflow and precipitation...

Table 5. Water-weight percent, organic-matter content, and dry-bulk ...

Table 6. Minor- and trace-element concentrations and particle size f...

Conversion Factors, Vertical Datum, and Abbreviated Units of Measurement

Temperature in degrees Fahrenheit (°F) can be converted to degrees Celsius (°C) as follows:

°C = (°F-32)/1.8

Vertical coordinate information is referenced to the National Geodetic Vertical Datum of 1929 (NGVD 29).

Altitude, as used in this report, refers to distance above the vertical datum.

Concentrations of minor and trace elements in sediment are given either in percent or micrograms per gram (mg/g).

Densities are given either as pounds per cubic foot (lb/ft³) or grams per cubic centimeter (g/cm³).

Abstract

Historical trends in streamflow, sedimentation, and sediment chemistry of the Wolf River were examined for a 6-mile reach that flows through the southern part of the Menominee Indian Reservation and the northern part of Shawano County, Wis. Trends were examined in the context of effects from dams, climate, and land-cover change. Annual flood peaks and mean monthly flow for the Wolf River were examined for 1907–96 and compared to mean annual and mean monthly precipitation. Analysis of trends in sedimentation (from before about 1850 through 1999) involved collection of cores and elevation data along nine valley transects spanning the Wolf River channel, flood plain, and backwater and impounded areas; radioisotope analyses of impounded sediment cores; and analysis of General Land Office Survey Notes (1853–91). Trends in sediment chemistry were examined by analyzing samples from an impoundment core for minor and trace elements. Annual flood peaks for the Wolf River decreased during 1907–49 but increased during 1950–96, most likely reflecting general changes in upper-atmospheric circulation patterns from more zonal before 1950 to more meridional after 1950. The decrease in flood peaks during 1907–49 may also, in part, be due to forest regrowth. Mean monthly streamflow during 1912–96 increased for the months of February and March but decreased for June and July, suggesting that spring snowmelt occurs earlier in the season than it did in the past. Decreases in early summer flows may be a reflection earlier spring snowmelt and large rainstorms in early spring rather than early summer. These trends also may reflect upper-atmospheric circulation patterns. The Balsam Row Dam impoundment contains up to 10 feet of organic-rich silty clay and has lost much of its storage capacity. Fine sediment has accumulated for 1.8 miles upstream from the Balsam Row Dam. Historical average linear and mass sedimentation rates in the Balsam Row impoundment were 0.09 feet per year and 1.15 pounds per square foot per year for 1927–62 and 0.10 feet per year and 1.04 pounds per square foot per year for 1963–99. Sedimentation in the impoundment was episodic and was associated with large floods, especially the flood-related failure of the Keshena Falls Dam in 1972 and a large flood in 1973. Sand deposition is common in the Wolf River upstream from the impounded reach for 2.5 miles and is caused by the base-level increase associated with the Balsam Row Dam. Some sand deposition also may have been associated with logging and log drives in the late 1800s and the failure of the Keshena Falls Dam. In the upstream 1.5-mile part of the studied reach, the substrate is mainly rocky; however, about 2,000 feet downstream from Keshena Falls, the channel has narrowed and incised since the 1890s, likely related to human alterations associated with logging, log drives, and (or) changes in hydraulics and sediment characteristics associated with completion of the Keshena Falls Dam and head race in 1908. Minor- and trace-element concentrations in sediment from Balsam Row impoundment and other depositional areas along the Wolf River generally reflect background conditions as affected by watershed geology and historical inputs from regional and local atmospheric deposition.

Introduction

The Wolf River is a recreationally and economically important natural resource for the Menominee Indian Tribe of Wisconsin. A study was begun in 1998 by the U.S. Geological Survey (USGS), in cooperation with the Menominee Indian Tribe, to examine historical trends in streamflow, sedimentation, and sediment chemistry of a 6-mi reach of the Wolf River as it flows through the southern part of the Menominee Indian Reservation and northern part of Shawano County, Wis. (fig. 1). For 24 mi upstream from Keshena Falls, the Wolf River was designated a Federal Wild and Scenic River in 1968. The State of Wisconsin also designated the river an Outstanding Resource Water in 1988.

Figure 1. Location of valley transects, stream gages, and precipitation sites for the Wolf River, Wis., study area.

Two dams, the Balsam Row Dam and the failed dam at Keshena Falls, form the lower and upper boundary of the studied reach and have affected channel characteristics in the reach (figs. 1–2). The Keshena Falls Dam failed in 1972 (fig. 2C). Anecdotal evidence from local residents suggests that sedimentation rates in the impoundment above the Balsam Row Dam and sand deposition in the free-flowing reach upstream of the impoundment have accelerated in the last three decades, possibly coincident with the failure of the Keshena Falls Dam. Flooding-related ice jams in winter also have been a problem in the town of Keshena after the failure of the Keshena Falls Dam (U.S. Army Corps of Engineers, 1995).

A

B

C

Figure 2. Photos of A, Balsam Row Dam in 1998, looking upstream at dam, B, Keshena Falls Dam in 1950, and C, failed Keshena Falls Dam, Wis., in 1972. (Photos B and C from the Wisconsin Department of Natural Resources, Dam File No. 72.1, Madison, Wis.)

The goals of this study were to (1) describe historical trends (for the last 100 years or more) in streamflow, sedimentation, and sediment-chemistry characteristics for the Wolf River and (2) identify major factors affecting flooding, sedimentation, and sediment-chemistry characteristics. Data collected as part of this study were useful for determining long-term and upstream/downstream effects from dams on channel characteristics of the Wolf River. Data are useful for establishing baseline conditions in case of future changes in hydrology, sedimentation, or sediment chemistry. Results from the study also provide information for protection or management of available resources, including water quantity, water quality, and aquatic habitat.

Purpose and Scope

This report presents the results from the study of historical trends in streamflow, sedimentation, and sediment chemistry for a 6-mi reach of the Wolf River between the Balsam Row Dam in Shawano County and Keshena Falls in the Menominee Reservation (fig. 1). Trends are described for annual flood peaks and mean monthly flow for the Wolf River based on about 90 years of combined data (1907–96) from two USGS stream gages—Keshena Falls and the Balsam Row Dam (near Shawano). Annual flood peaks during 1968–96 from the stream gage near Shawano were compared to flood peaks at an upstream gage on the Wolf River near Langlade, Wis. Precipitation data from three nearby precipitation stations for 1907–96 are compared to the streamflow trends. Sedimentation patterns in channel, backwater, flood-plain, and impounded areas, historical sedimentation rates, and channel changes are described for a period from about 1850 (pre-European settlement) to 1999. Lastly, minor- and trace-element concentrations in impounded, backwater, and flood-plain sediment are described in terms of historical trends (about 1850–1999) and spatial distribution. This report attempts to evaluate the effects of climate change, logging, forest regrowth, and dams on trends in streamflow, sedimentation, and sediment chemistry over the historical period of interest.

Description of Study Area

The Wolf River upstream of the Balsam Row Dam has a drainage area of 816 mi² and has a temperate, continental climate characterized by large seasonal changes in temperature but little variation in monthly precipitation. The Wolf River Basin is typified by an average annual temperature of 43°F, about 30 in. of annual precipitation, 48 in. of annual snowfall, and 18 in. of annual evapotranspiration (Wendland and others, 1985; Olcott, 1968).

The reach of the Wolf River included in this study (fig. 1) generally flows south through coarse-grained glacial deposits and outwash (Clayton and others, 1991; Milfred and others, 1967). These deposits are underlain by Precambrian granite and syenite rocks associated with the Wolf River Complex (Mudrey and others, 1982). Rocks associated with the Wolf River Complex crop out at Keshena Falls but are buried by glacial, outwash, and fluvial deposits to the south and through most of the study reach.

Flood-plain soils near the Wolf River are characterized as loamy Fluvaquents or Fluvents and are underlain by coarse- and fine-grained deposits associated with outwash plains or glacial lakes (Gundlach and others, 1982). The flood-plain soils generally contain about an 8-in. thick organic surface layer, although some have as much as 15 in. of organic-rich loam at the surface. Upland soils near the studied reach are well-drained sandy or loamy Spodosols or Alfisols (Gundlach and others, 1982).

The slope of the Wolf River is highly variable through the Menominee Indian Reservation but flattens out considerably below Keshena Falls (fig. 3). Based on measurements of stream length and contour intervals from USGS 7.5-minute topographic maps, the average slope is 11.6 ft/mi from the headwaters to Big Smoky Falls, and the average slope from Big Smoky Falls to Keshena Falls is 4.9 ft/mi. In contrast, the average slope is 2.0 ft/mi downstream from Keshena Falls to the Balsam Row Dam.

Figure 3. Longitudinal profile of the Wolf River through the Menominee Indian Reservation, Wis.

Land cover in the Wolf River Basin upstream from the Balsam Row Dam is mainly forest (72 percent of the basin) and wetland (15 percent) as inferred from 1992–93 satellite imagery (Reese and others, 2002). Cropland accounts for only 4 percent of the drainage area. Of the forested land, 78 percent is deciduous (mainly aspen and maple), 16 percent is mixed deciduous and conifer, and 6 percent is conifer. Along the study reach, vegetation in the upland generally is categorized as pine plantation (Milfred and others, 1967). The Menominee Indian Tribe has used selective logging and sustainable forestry techniques on their lands since the reservation was established in 1854 (Menominee Tribal Enterprises, 1997). The reservation covers the lower third of the Wolf River Basin (fig. 1). Most of the upper part of the basin was logged before 1900 (Connor, 1978). The Wolf River and its tributaries were used for transportation of saw logs from the upper basin to Shawano and further downstream during the late 1800s (Oehmcke and Truax, 1964; Connor, 1978). Logging in the area peaked in 1873.

In the late 1800s the upper Wolf River had various small dams related to logging and log drives (Smith, 1908; U.S. Department of Agriculture and the University of Wisconsin, 1961). Both the Keshena Falls Dam and the Balsam Row Dam were used for hydroelectric power generation. Neither dam has flood-control function. The presence of some sort of dam structure at Keshena Falls dates back to the 1870s and was related to logging, but the most recent dam was constructed in 1906–08 and failed in 1972 (Wisconsin Department of Natural Resources, Dam File No. 72.1, Madison, Wis.). The Keshena Falls Dam had about 16 ft of head. The dam consisted of two parts—a wooden apron that extended across the main channel of the Wolf River and four gates across a diversion canal (head race) for the powerplant (fig. 2B). The failure of the Keshena Falls Dam in 1972 happened after about 9 years of disrepair along the left side of the wooden spillway apron (Wisconsin Department of Natural Resources, Dam File No. 72.1, Madison, Wis.) (fig. 2C).

The Balsam Row Dam (fig. 2A) was constructed in 1927 and there is no indication of any earlier version of a dam at Balsam Row before then (Smith, 1908). The Balsam Row Dam has six gates and about 14 ft of head. The powerhouse is on the east side of the dam embankment next to the gates.

A general description of the surface-water and ground-water quality for the Menominee Indian Reservation was previously published in Krohelski and others (1994). Garn and others (2001) provide additional data on the quality of water, sediment, and selected benthic organisms from the Wolf River during 1986–98.

Methods

This study involved collecting and compiling historical records as well as collecting new field data. General Land Office Survey notes provided information on channel characteristics before and during European settlement, logging, and log drives in the mid to late 1800s.

Streamflow trend analyses involved statistical examination of trends in annual peak and average monthly flow data from a former USGS stream gage at the Balsam Row Dam near Shawano, Wis. (USGS station number 04077400), with 93 years of data (1907–09, 1910–2001). The stream gage was moved from Keshena Falls to the Balsam Row Dam in 1985 (table 1). Mean daily flow data were examined before and after the site location changed to confirm that streamflow characteristics were the same at both locations. Annual peak data from the Shawano site also were compared to data from the Wolf River near Langlade, Wis. (USGS station number 04074950). The drainage area of the Wolf River at Langlade is about half that of the Wolf River near Shawano (table 1). Trend analysis was also done on annual and monthly precipitation from three nearby National Weather Service climate stations (Rhinelander, Shawano, and Green Bay, Wis.) over the same period to identify climatic effects on streamflow characteristics.

Table 1. U.S. Geological Survey stream gages used in the analyses of streamflow trends from the Wolf River, Wis.

[mi², square miles]

Analysis of historical changes in sedimentation and sediment quality involved collection of cores and elevation data along nine valley transects spanning the Wolf River channel, flood plain, and impounded area above the Balsam Row Dam from February through June 1999 (table 2). Locations of transects are shown in figure 1. The number of cores collected along each transect varied because of differences in the number of depositional environments found along each transect. For example, transect T4.2 was along a stretch of the Wolf River with no flood plain and a uniform sand channel bottom. In contrast, transect T7 intersected a longitudinal bar, a backwater channel, and flood-plain deposits and required more cores for an adequate description of the range of fluvial environments traversed. Sediment samples were selected from cores of the various environments and analyzed for sediment chemistry, radiometric age dating, and particle-size distribution. In November 2001, a second core was collected near core T2-1 to obtain additional information on water, organic-matter content, and dry-bulk density.

Table 2. Description of valley transects and cores collected along the Wolf River from the Balsam Row Dam to Keshena Falls, Wis.

[Distances are curvilinear channel distances; ft, feet]

¹ No valley transect surveys were done for T5 and T6.

Analysis of General Land Office Survey Notes

General Land Office (GLO) Surveys were done in the mid 1800s across Wisconsin to map and establish townships, ranges, and sections. In northern Wisconsin, these surveys usually were done before European settlement and were useful for characterizing stream conditions before widespread logging and burning coincident with European settlement in the late 1800s. Copies of original GLO survey notes were obtained from the State of Wisconsin, Board of Commissioners of Public Lands for two townships: Township 27 North, Range 15 East, and Township 28 North, Range 15 East. The surveys were done from 1845 through 1891, with notes for the south boundary of Township 28 from 1845, Township 27 from 1853, and Township 28 from 1891.

Field notes from the GLO Surveys contain information on vegetation, soils, and the lay of the land, as well as the location, width, depth, substrate, swiftness of flow, and direction of streams at section-line crossings. These notes were used to determine the location of the Wolf River channel before the impoundment of water behind the Balsam Row Dam. Meander data for the Wolf River also were reported; these data described the location of both streambanks longitudinally.

Streamflow and Precipitation Data Analyses

Data from two USGS stream gages on the Wolf River were examined for trends in annual flood peaks and mean monthly flow: the Wolf River near Shawano (the Balsam Row Dam) and the Wolf River near Langlade (fig. 1, table 1). The gage near Shawano was moved from Keshena Falls to the Balsam Row Dam in 1985 and was operated at the Balsam Row Dam until 2001. Streamflow data for the two gages were combined and recorded under site 04077400 in USGS databases because of similar drainage areas (table 1). Annual flood peaks for the combined record are available for 1907–09 and 1910–96. Mean monthly data are available for 1912–96. Mean daily flow and annual peaks from 1975–85 and 1985–95 were examined for differences by use of two-sample t-tests with log-transformed data and the nonparametric Wilcoxon rank sum test (Iman and Conover, 1983). Gebert and Krug (1996) also combined data from the Balsam Row Dam and Keshena Falls gages and examined trends in annual flood peaks and low flow for the period of record 1931–91.

Tests for normal distributions of the data were performed and log transformations were used to normalize the data. The log-transformed data were analyzed for trends by use of the least squares method for linear regression (Iman and Conover, 1983). Slopes in the trend lines were considered statistically significant if p-values were less than 0.10. Trend analyses were done on annual peak flow and mean monthly data for 1907–96 and also divided into pre- and post-1950 periods because hemispheric-scale climate change is thought to have occurred around 1950 (Knox and others, 1975; Kalnicky, 1974). Although highly variable and season-dependent, upper atmospheric circulation patterns generally changed from more zonal (west-east) before 1950 to more meridional (north-south) after 1950. Changes in upper atmospheric circulation in Wisconsin affect the frequency, intensity, and seasonality of precipitation (Knox and others, 1975). Meridional patterns tend to produce more floods on an annual basis than any other circulation pattern. Zonal patterns tend to produce mild weather in the Great Lakes region (Eichenlaub, 1979).

Annual precipitation data from three long-term climate stations (Rhinelander, Shawano, and Green Bay, fig. 1) were obtained from the State Climatologist's Office in Madison, Wisconsin. The stations are operated by the National Weather Service. The Rhinelander station had data beginning in 1908. Precipitation records for Shawano began in 1923 and for Green Bay began in 1912. Trend analyses (least squares method for linear regression) were done on average annual and average monthly precipitation data in the same fashion as was done with the streamflow data.

Transect Surveys and Sediment Coring

Data collected from valley transect surveys and sediment coring were used to determine historical sedimentation rates and historical trends in sediment chemistry. A reconnaissance trip was done in October 1998 along the Wolf River from Keshena Falls to the Balsam Row Dam to identify possible transects for sediment coring. Nine transects (fig. 1; table 2) were selected along the study reach and numbered in upstream order, the starting point being at the Balsam Row Dam. Transects were located in reaches with a flood plain and with depositional zones within the channel. Transects were surveyed in May and June 1999 by use of an auto level. A global positioning system was used to verify the location of each transect. Elevation data were tied into U.S. Army Corps of Engineers reference marks from a flood-plain delineation study done in 1995 (U.S. Army Corps of Engineers, 1995). Soft- and hard-bottom elevations for transect T1b, T1c, T2, T3, and T4 were collected by use of a sounding pole in unwadeable areas. Hard-bottom elevations are assumed to be elevations of pre-dam sediment surfaces (streambed, streambank, backwater, or flood plain). No valley-transect surveys were done at T5 and T6; only exploratory cores from channel deposits were collected during the reconnaissance trip.

The thickness, texture, chemistry, and age of sediment in the flood plain, backwater areas, impoundment, and modern channel were examined by sediment coring and sampling with four devices, including a 1-in.-diameter hand-held soil probe, a geoprobe, and two types of piston corers—the Wisconsin Department of Natural Resources (WDNR) corer (3-in. diameter) and the modified Livingston corer (2-in. diameter) (table 3). The hand-held soil probe was useful for quick exploratory coring of flood-plain deposits and field sediment descriptions. The geoprobe is a tapping-type coring device that was mounted on a four-wheel all-terrain vehicle and was used to collect cores from the impounded reach when it was frozen and from accessible flood-plain areas. The WDNR corer is limited to depths less than about 2 m and was used to collect cores from inundated areas. The modified Livingston was used to collect multiple cores at depth from fine-grained deposits (Wright, 1967).

Table 3. Summary of core locations, water depths, penetration depths, samples, and types of cores collected from the Wolf River, Wis.

[ft, feet; LB, left bank; WDNR, Wisconsin Department of Natural Resources; te, minor and trace elements; cs, ¹³⁷Cs and ²¹⁰Pb radiometric dating; ps, particle size; RP1, Reference point (survey stake) for transect endpoint; LEW, left edge of water; u.s., upstream; wd, radiocarbon; --, not measured]

Cores were collected along transects T1b, T1c, and T2 with the geoprobe and the piston corers in February and March 1999 while the impoundment was frozen. Elevations of pre-dam surfaces identified in the cores were compared to elevations of the hard bottom determined with the sounding pole. The remainder of the coring was done in May and June 1999 by use of a boat or by wading. Backwater, channel, and impoundment deposits were cored with the WDNR and Livingston piston corer. The WDNR corer was used first, and then the Livingston corer was used to collect cores at deeper intervals. Before coring, potential sites were submitted to the Environmental Services Department, Menominee Indian Tribe for checking against known archeological sites. No cores were collected from known archeological sites.

All cores were described for texture and color in the field by use of the U.S. Department of Agriculture textural triangle and color chart (Munsell Color, 1975; Soil Survey Staff, 1951). Field grading of texture was done by rubbing soil between the fingers (Milfred, 1967). Recovery ratios were recorded for all coring devices. Sediment samples from a subset of cores were analyzed for water content, organic matter, particle size, chemistry, and radiometric dating (table 2). County-level soil survey data also were used to verify the lateral extent of deposits observed in the cores (unpublished Menominee County Soil Survey, about 1999; Gundlach and others, 1982).

Buried soils are commonly found in flood-plain deposits and can be good indicators of flood-plain surface stability. Buried soils represent older flood-plain surfaces (Birkland, 1984; Retallick, 1985). In general, modern flood-plain soils are usually poorly developed because of the possibility of a fast rate of burial, high water table, and anthropogenic disturbance. Many buried flood-plain surfaces will have the beginnings of an A horizon or only a thin layer of organic matter and remnants of vegetation. The A horizon is a dark zone at the surface of the soil caused by accumulation of decomposing organic matter. The buried flood-plain surfaces can be recognized by one or more of the following characteristics: presence of an A horizon, organic matter accumulation from flood-plain vegetation, lateral extent (a buried surface is parallel to the land surface and may truncate geologic bedding), root traces, and soil structure (Birkland, 1984; Retallick, 1985). Sometimes, part of a buried soil will be removed by scour activity associated with flooding, channel migration, or channel avulsion.

Laboratory Analysis

Physical Characteristics and Organic-Matter Content

Physical characteristics measured included water-weight percent, particle size, and organic-matter content (loss on ignition). Water-weight percent and organic-matter content were analyzed at the USGS, Middleton, Wis., for 47 samples collected from core T2-1. Water-weight percent was measured by use of standard American Society for Testing and Materials Procedure D2216-92 (1992), except that sample sizes were less than 20 g wet weight because of the numerous samples from the single core. Water-weight percent was determined by measuring weight loss after 24 hours at 105°C. Organic-matter content was determined by weight loss after ashing (loss on ignition, or LOI) at 550°C for 1 hour (Dean, 1974). Dry-bulk density was estimated by use of the formula

where ρ is dry-bulk density (g/cm³), x is depth in the core, D is proportion dry weight of unit wet volume, I is inorganic proportion of dry material (assuming a density of 2.5 g/cm³), and C is organic proportion of dry material (assuming a density of 1.6 g/cm³) (HÅkanson and Jansson, 1983). The density of water is assumed to be 1.00 g/cm.

Twenty-six samples were analyzed for particle size at the USGS Cascades Volcano Laboratory at Vancouver, Wash. Samples were analyzed for percent sand, silt (four size fractions), and clay by the sieve-pipet method (Guy, 1969).

Radiometric Dating and Sedimentation Rates

Analyses of ¹³⁷Cs, ²¹⁰Pb, and ²²⁶Ra were done on twelve 3-in. intervals from core T2-1, which contained 84 in. (7 ft) of post-dam sediment, and eight 0.1-ft intervals from core T7-2, which contained 1 ft of post-dam sediment. Analyses were done at Quanterra Analytical Services, Richland, Wash. Activity concentrations of ²¹⁰Pb, ²²⁶Ra, and ¹³⁷Cs were determined from gamma spectroscopy counting. The profile for ¹³⁷Cs is used to compare sedimentation rates before and after 1963. In 1945, ¹³⁷Cs was first detected, and in 1954 the first increase occurred in the Northern Hemisphere, corresponding to increased nuclear weapons testing (Krishnaswami and Lal, 1978). In 1960 a minimum occurred, followed by a maximum in 1963. With the signing of the atmospheric nuclear test ban treaty, atmospheric contributions have dropped off substantially (Olsson, 1986). A date of 1963 was assigned to the sample with the highest ¹³⁷Cs activity. Mass and linear sedimentation rates were determined for 1927–63 and 1963–99, from dam construction to the time the cores were collected.

The ²¹⁰Pb dating technique is based on the escape of radon from the Earth and the subsequent decay of this radioactive gas into ²¹⁰Pb. This technique is most suitable for dating within the last 150 years because the half-life of ²¹⁰Pb is about 22 years (Olsson, 1986). To account for some of the variability associated with possible fluctuation in the sources of lead or inhomogeneous sediment, ²²⁶Ra also was analyzed in each of the samples. Optimally, samples from sediment more than 150 years old are collected to measure local background concentrations of ²¹⁰Pb supplied to the sediment from decay of uranium minerals (Olsson, 1986). Unsupported (excess) ²¹⁰Pb activity was estimated by subtracting the average 226Ra activity for all sampled intervals in a core from total ²¹⁰Pb activity. For the constant initial concentration model, the amount of unsupported ²¹⁰Pb activity in the uppermost sediment layer is assumed to be constant through time; thus, a profile of ²¹⁰Pb activity exhibits a log-linear decrease with depth (Goldberg, 1963; Krishnaswami and others, 1971; and Robbins, 1978).

Wood or organic material was collected from backwater and impoundment cores from transects T4, T7, and T9 and nine samples were radiocarbon dated (¹⁴C). Radiocarbon analyses can be useful for dating carbon-rich samples from about 200 to 40,000 years old (Bowman, 1990; Libby, 1946). Two types of radiocarbon analyses were done. Two samples were large enough for conventional radiocarbon techniques, which use beta particle counting from decaying ¹⁴ C atoms to estimate the age of the sample. Seven samples weighed less than 3 g and required the use of accelerator mass spectrometry (AMS). All samples were analyzed at Beta Analytic, Inc., Miami, Fla. Calendar calibrated age ranges were calibrated by the Pretoria calibration procedure (Talma and Vogel, 1993; Stuiver and others, 1998).

Sediment Chemistry

Twenty-eight samples from six cores were submitted to the USGS National Water-Quality Laboratory, Denver, Colo., to be analyzed for 26 minor and trace elements. Plastic equipment (Teflon, polypropylene, or polyethylene) was used to subsample cores. The WDNR piston core tube was composed of Lexane, and the geoprobe cores had plastic liners. The Livingston core tube was composed of stainless steel. Samples were not sieved. Laboratory methodology is detailed in Arbogast (1996) and Briggs and Meier (1999). Total digestions were done.

Quality Control

Quality-control procedures included collecting and analyzing samples from duplicate cores and about 15 percent replicate samples to assess within-site and analytical variance. Quality-control measures at the USGS laboratories included comparisons to standard reference materials, spikes, and duplicates (Pirkey and Glodt, 1998).

Streamflow Trends

Results from two-sided t-tests indicate no statistically significant differences for streamflow between the Balsam Row Dam and Keshena Falls stream gages for either mean daily flows or annual peaks over the period 1975–96. Wilcoxon signed-rank tests for the same data yielded similar results. Boxplots illustrate the absence of difference between the sites in mean daily flow and annual flood peaks (fig. 4). These results indicate that combining data from both sites for analyses of streamflow trends was appropriate. Hereafter, streamflow trends from the combined records are referred to as the record from the Wolf River near Shawano.

Figure 4. Boxplots of A, mean daily streamflow, and B, annual peak floods for the Wolf River near Shawano (Balsam Row Dam), Wis., and the Wolf River at Keshena Falls, Wis. The stream gage was moved from Keshena Falls to Balsam Row Dam in 1985.

Annual flood peaks from the Wolf River near Shawano for the entire period of record (1907–96) showed no statistically significant trend based on linear regression on log-transformed discharge data (table 4; fig. 5A). These results are similar to those found by Gebert and Krug (1996) for the period 1931–91. However, when the annual flood peaks are divided into two periods before and after 1950, there is a statistically significant downtrend in annual flood peaks for 1907–49 and an uptrend for 1950–96 (fig. 5B). During 1907–49, only three large floods with a recurrence interval of about 25 years occurred—those in 1912, 1922, and 1929. (The recurrence interval is based on the entire record from 1907–96.) The 1912 flood was in September and the 1922 and 1929 floods were in April. During 1950–96, four large floods occurred with recurrence intervals of 50–100 years: those in 1960 (75-year), 1973 (greater than 100-year), 1976 (50-year), and 1986 (50-year). These floods were in March–May and resulted from a combination of snowmelt and intense rainfall. In general, these trends are similar to results found for floods in the Upper Mississippi Valley, where upper atmospheric circulation patterns were mainly zonal before 1950 and meridional after 1950 (Knox and others, 1975). Meridional (north-south) patterns tend to produce more floods on an annual basis than other circulation types.

[A negative t-statistic reflects a decreasing trend and a positive t-statistic reflects an increasing trend. Statistically significant trends (p-values < 0.10) are in bold.]

Figure 5. Annual peak streamflow from the Wolf River near Shawano (the Balsam Row Dam), Wis. for 1907–1996: A, trend line for entire period of record, and B, trend line for 1907–49 and 1950–96.

Data collection at the Langlade stream gage on the Wolf River began in 1968, much later than at the Shawano gage (table 1). A comparison of annual flood peaks from Langlade and Shawano for 1968–96 showed no significant trend in annual flood peaks for the period (fig. 6). The records are somewhat similar, but large floods in the lower Wolf (Shawano) did not always correspond to large floods in the upper Wolf (Langlade). Four of the six major floods above 3,000 ft³/s at Shawano (1973, 1976, 1986, and 1996) corresponded to similarly large floods at Langlade (fig. 6). Examination of the time of year for the floods indicates that spring snowmelt coupled possibly with frontal-type precipitation (large extent, low intensity, long duration) caused large floods in both the upper and lower Wolf River. However, convectional precipitation (summer thunderstorms with limited extent, high intensity, and short duration) resulted in more localized flooding, as indicated by June floods in 1969 and 1993 for the lower Wolf River but not the upper Wolf River.

Figure 6. Annual peak flow and trend lines for the Wolf River near Shawano, Wis., and the Wolf River at Langlade, Wis., 1968–96.

Similar to annual flood peaks, annual precipitation data for the entire period of record did not show any significant trends (p > 0.10) for the three precipitation sites (table 4). For 1912–49, the Green Bay site had a downtrend in annual precipitation (fig. 7), similar to the downtrend in annual flood peaks (fig 5B). However, data from the Rhinelander site had an uptrend that was not statistically significant, and the Shawano site had a nonsignificant downtrend (table 4). The Rhinelander and Shawano sites are closer to the Wolf River Basin than the Green Bay site is (fig. 1). Annual precipitation data from all three sites showed a nonsignificant uptrend during 1950–96 (table 4, fig. 7). Therefore, trends in annual flood peaks for the Wolf River may be related to trends in annual precipitation, but the relations are confounded. In general, zonal circulation patterns tended to decrease precipitation, whereas meridional circulation patterns tended to increase precipitation in the upper Mississippi Valley (Knox and others, 1975). Perhaps annual precipitation data are not sensitive enough to determine frequency of rainstorms that produce large floods, especially spring floods that are related to a combination of snowmelt and rainstorms.

Figure 7. Annual precipitation and trend lines for Green Bay, Wis., 1912–49 and 1950–96.

Trends in mean monthly flow at the Wolf River near Shawano were observed for some months for the entire record (1912–96) and during 1912–49. During 1912–96, mean monthly flow for February and March increased, whereas mean monthly flow for June and July decreased (table 4). These results also were found during 1912–90 for the same gage (Peters, 1997). July mean streamflow also decreased significantly during 1912–49 (table 4).

In contrast to the trends at the Wolf River near Shawano, no trends in monthly precipitation at Green Bay were found for March, June, or July for 1912–96, and February precipitation for 1912–96 decreased. Only decreasing July precipitation during 1912–49 matched decreasing July mean monthly flow during 1912–49 (table 4). Therefore, the uptrend during 1912–96 in mean monthly flow for February and March indicates that snowmelt is occurring earlier. An earlier snowmelt also may cause an apparent decrease in streamflow during the early summer months because low-flow conditions may start earlier in the summer season. The downtrend in July precipitation and streamflow during 1912–49 may indicate a decrease in the occurrence of summer storms during that period.

A secondary cause for decreases in early summer monthly flow over the period of record may be related to forest regrowth. Most of the old growth forests of northeastern Wisconsin were logged and burned after European settlement in the late 1800s. Widespread logging and burning continued in the region until the late 1920s (Connor, 1978; Fries, 1951), although the Menominee Indian Tribe practiced selective cutting techniques on their land since 1854 (Menominee Tribal Enterprises, 1997). In northern Wisconsin, property taxes were based on the amount of board feet per acre up until 1927, which promoted repeated cutting and burning after the initial logging. It is worthwhile to include a description of logging in the Wolf River Basin as described by Samuel Shaw on December 9, 1886, in the "Forest Republican". He states, "The Upper Wolf River is stripped clean of pine. The Lower Wolf River is bare. There is nothing to tempt a lumberman to lift an ax or draw a saw" (Connor, 1978). By the late 1970s, second- and third-generation forest covered much of the Upper Wolf River Basin (Connor, 1978). Much of the land in the basin in 1999 was owned by the Menominee Tribe, the U.S. Forest Service, counties, or private lumber companies.

Flood peaks would be expected to decrease during forest regrowth as surface runoff is decreased or slowed because of increases in interception, hydraulic roughness, and infiltration (Fitzpatrick and others, 1999). Base flow also would be expected to decrease because of an increase in evapotranspiration, especially during summer. Forest evapotranspiration reduces the amount of water available for ground-water recharge and base flow (Sartz, 1972).

Annual flood peaks and runoff volumes from three watersheds near the Wolf River with similar logging histories also have shown relations to both precipitation and forest-cover change (Fitzpatrick, 1993; Rayburg, 1999). The Pike and Oconto Rivers are to the east of the Wolf River (fig. 1), and annual flood peaks and runoff volumes for those streams decreased starting in about 1930 (Fitzpatrick, 1993). The Pike River Basin (257 mi ² ) is smaller than the Wolf River Basin and is directly east of the Rhinelander climate station. However, no downtrend was observed in annual precipitation at Rhinelander for 1912–49, suggesting that streamflow in the Pike River was more likely affected by forest regrowth than by precipitation. In contrast, trends in annual flood peaks and runoff volumes for the Oconto River Basin (737 mi ² ) corresponded more closely to precipitation at the Oconto precipitation site. Historical streamflow trends in the Prairie River Basin (181 mi2), about 50 mi northwest of the Wolf River study reach, primarily were attributed to land cover and secondarily to climate change (Rayburg, 1999). Mean monthly flow increased in the Prairie River during the winter, but total annual flow decreased. Reduction in flow volumes occurred in the Prairie River during the growing season.

In summary, three large floods occurred on the Wolf River near Shawano between 1912 and 1929, followed by no large floods during 1930–59. Four large floods occurred between 1960 and 1986. The trends toward smaller annual flood peaks during 1908–49 compared to larger annual flood peaks during 1950–96 appear to be related to precipitation and upper atmospheric circulation patterns. The trend toward smaller annual flood peaks during 1908–49 may secondarily be related to forest regrowth. During 1908–96, floods became more common in early spring, and mean monthly flows decreased in June and July. The flood records for the lower Wolf River (near Shawano) and the upper Wolf River (Langlade) are similar, but large floods in the lower Wolf did not always correspond to large floods in the upper Wolf. Snowmelt coupled with frontal-type precipitation caused large floods in both parts of the basin, whereas floods from intense summer rainstorms were more localized.

Sedimentation Trends

Coring, sounding, and survey data from nine transects provided the information necessary to reconstruct the sedimentation history of the Wolf River through the study reach. Transect locations and number of cores collected at each transect are listed in table 2, and water depth, penetration depth, and type of sample analyses for the cores collected along each transect are summarized in table 3.

The GLO notes (from 1853 for the lower part) describe the reach as having a swift current and in places rapids. Banks were generally high. The channel bottom was composed of sand and rocks, and water depths were from 1 to 3 ft. The river was reported to be navigable for steamboats at high stages (Oehmcke and Truax, 1964). The GLO notes describe the presence of rapids or falls just below the county line with a 10- to 12-ft drop over 400 ft of channel length. Evidence of the rapids has disappeared after that part of the reach was covered with impounded water and sediment associated with the Balsam Row Dam. The GLO notes also provided data on the location and planform of the Wolf River through the study reach (fig. 8). These data were compared to the coring and sounding data and used to confirm the pre-dam location of the Wolf River in the impounded reach.

Figure 8. Trace of the Wolf River, Wis., showing channel position in 1853 (before impoundment by the Balsam Row Dam).

A detailed longitudinal profile of the Wolf River thalweg (deepest part of the channel) in 1999 and before the construction of the Balsam Row Dam was created from the 1999 coring, sounding, and survey data from the nine valley transects and supplemented with information from the GLO notes, 1998 exploratory cores from the two sites (T5 and T6) with no valley cross-section surveys (table 3), and data from the 1995 U.S. Army Corps of Engineers (USACE) flood-plain delineation study (U.S. Army Corps of Engineers, 1995). Distances along the longitudinal profile (fig. 9) are curvilinear channel distances; these are the same distances used by the USACE study, which are listed in table 2 along with the upstream distance of each transect from the Balsam Row Dam.

Figure 9. Detailed longitudinal profile of the Wolf River from the Balsam Row Dam to Keshena Falls, Wis., showing pre-dam thalweg, 1999 thalweg, and approximate water surface at low flow.

The longitudinal profile was divided into three reaches on the basis of sedimentation history and streambed characteristics. The impounded reach extends from the dam to the county line and encompasses the part of the study reach with silty clay deposition. The sandy reach, from the county line to the Fairground Road Bridge, encompasses the part that has noticeable sand deposition. The upper reach, from the Fairground Road Bridge to Keshena Falls, has predominantly rocky/gravel substrate and little deposition of sand, silt, or clay.

The following three-part description of sedimentation history for the study reach is based on the above substrate conditions. This description proceeds from the downstream boundary to the upstream boundary of the study reach. Geologic sections are shown in figures 10–12 and in figures 15–20 for each transect looking south (downstream).

Impounded Reach

The impounded reach extends from about 1,400 to 11,000 ft (1.8 mi) along the longitudinal profile, from the Balsam Row Dam to the county/reservation boundary (fig. 9). About 5 to 10 ft of organic-rich silty clay (muck) was deposited in the impoundment from 1927 to 1999. The normal water-surface elevation at the dam is about 15 ft above its pre-dam elevation. The bulk of the sediment was deposited in the 9,000-ft reach between the dam and the buried falls downstream of the county line. The impounded reach varies in width from about 400 to 700 ft.

Transects T1b, T1c, and T2 represent typical conditions in the impounded reach (figs. 10–12). Post-dam deposits are composed of organic-rich stratified clay and silty clay, with interbedded layers of abundant woody debris, leaf debris, pine needles, and rootlets. These layers of organic debris are thought to represent deposits from large floods. As stated earlier, five floods with a recurrence interval of 25 years (greater than 4,100 ft³/s) or more occurred after the dam was constructed in 1927 (fig. 5B). Four of the five floods were between 1960 and 1986, and two were within 4 years of each other (1973 and 1976). The layers of organic debris were common in the upper part of the impounded deposits and were especially noticeable in cores from transect T1b (fig. 10). There was less organic debris in the impounded sediment on the east side of the impoundment through its widest stretch at T2 (fig. 12). Most of the organic debris settled out in the narrow parts of the impoundment or toward the west side of the impoundment, where it widens (fig. 12).

Figure 10. Valley transect through the impounded reach of the Wolf River, Wis., T1b, at 1,850 feet along the longitudinal profile (fig. 9).

Figure 11. Valley transect through the impounded reach of the Wolf River, Wis., T1c, at 2,260 feet along the longitudinal profile (fig. 9).

Figure 12. Valley transect through the impounded reach of the Wolf River, Wis., T2, at 2,790 feet along the longitudinal profile (fig. 9).

The pre-dam (pre-1927) channel through the impounded reach was identified by the presence of a gravel lag deposit and relative elevation interpreted from core and sounding data. The pre-dam flood-plain surface was identified by the presence of a buried A horizon, abundant root remnants, grass remnants, and the presence of loam or sand. The 1927 channel was at the same location as the 1853 channel as reconstructed from the GLO notes (fig. 8). Upstream from the dam at transect T1b, the pre-dam channel was on the west side of the impoundment (fig. 10). Further upstream, at transects T1c and T2, the pre-dam channel was on the east side of the impoundment (figs. 11–12). In contrast, most of the flow throughout the length of the impoundment in 1999 was on the west side because the dam gates were usually open on the west side.

Profiles for ¹³⁷Cs and ²¹⁰Pb from core T2-1 were used to distinguish linear and mass sedimentation rates for the impoundment (fig. 13). The undisturbed ¹³⁷Cs profile from core T2-1 with a sharp peak indicates no bioturbation or postdepositional mixing in this core. The ¹³⁷Cs profile shows a sharp peak at 3.6 ft, which is assumed to correspond to the peak fallout of ¹³⁷Cs in 1963. The expected monotonic decay curve for the ²¹⁰Pb profile shows a break from about 2.6 to 3.1 ft. This break may be the input of older sediment stored behind the Keshena Falls Dam (1908–72) and transported to the Balsam Row impoundment after the failure of the Keshena Falls Dam. Based on core data, the 1927 surface was at 7.0 ft. Linear sedimentation rates, based on the ¹³⁷Cs peak and the 1927 surface, are similar before 1963 and after 1963, at about 0.09 ft/y and 0.10 ft/y, respectively.

Figure 13. Data for ¹³⁷Cs, unsupported ²¹⁰Pb, and ²²⁶Ra from core T2-1 from the Balsam Row Dam impoundment, Wolf River, Wis.

Data for particle size, dry-bulk density, and organic-carbon content from core T2-1 indicate that the source of sediment to the impoundment most likely changed over time and that deposition probably was episodic (fig. 14; table 5). Impounded sediment averaged about 51 percent clay, 43 percent silt, and 6 percent sand. An increase in sand at about 2.2 ft (mid-1970s) is followed by an increase in silt upcore from 2.2 to 1.7 ft. This shift in composition may reflect input of sediment from behind the Keshena Falls Dam that was likely transported downstream during the 1972 dam failure and the 1973 flood (largest flood on record). The slight increase in silt, increase in bulk density, and decrease in organic carbon at about 4.0 ft (about 1960) may reflect an increase in coarser sediment delivered during the 1960 flood, which was the first large flood to occur after 31 years of relatively small floods between 1929 and 1960 (fig. 5B). Also, during the 1960 flood, an earth-fill wall near the powerplant at the Keshena Falls dam was dynamited to relieve pressure on the dam (U.S. Geological Survey, 1961). This action also may have caused a pulse of more coarse-grained sediment to the Balsam Row impoundment. Low concentrations of organic carbon in sediment cores from impoundments in the Upper Mississippi River Basin also were attributed to large floods (Juracek, 2004).

Figure 14. Profiles for A, particle-size distribution, B, bulk density, and C, organic carbon concentration collected from a core at T2-1 in 2001 from the Balsam Row Dam impoundment, Wolf River, Wis.

Table 5. Water-weight percent, organic-matter content, and dry-bulk density for samples from Wolf River, Wis., sediment core T2-1, 2001.

[ft, feet; lb, pound; lb/ft³, pound per cubic foot]