17289 Clinton River Sed Transport.pdf

U.S. Army Corps of Engineers Detroit District

Clinton River Sediment Transport Modeling Study

Prepared by W.F. Baird & Associates Ltd.

Madison, Wisconsin

May 16, 2005

CLINTON RIVER SEDIMENT TRANSPORT

MODELING STUDY

Prepared for:

U.S. ARMY CORPS OF ENGINEERS

DETROIT DISTRICT

CONTRACT # DACW35-01-D-0009/0012

Prepared by:

W.F. BAIRD & ASSOCIATES LTD.

MADISON, WISCONSIN

MAY 16, 2005

USACE – Detroit District Clinton River Watershed Model Great Lakes Hydraulics and Hydrology Office Sediment Transport Modeling Study

Reporting History

Rev.

No

Date Status Comments Reviewed by Approved by

0 01-10-05 First Draft FGM BEH 1 01-24-05 Second Draft FGM BEH 3 05-10-05 Final FGM BEH

For further information please contact James P. Selegean, P.E, PhD.

(313) 226-6791

This report was prepared by W.F. Baird & Associates Ltd. for USACE Detroit District.

The material in it reflects the judgment of Baird & Associates in light of the information available to them at the time of preparation. Any use which a Third Party makes of this report, or any reliance on decisions to be made based on it, are the responsibility of such Third Parties. Baird & Associates accepts no responsibility for damages, if any, suffered by any Third Party as a result of decisions made or actions based on this report.

USACE – Detroit District Clinton River

TABLE OF CONTENTS

EXECUTIVE SUMMARY

1.0 INTRODUCTION

1.1 Purpose

1.2 Watershed Description

1.3 Clinton River Watershed Sediment Delivery

1.3.1 Terminology

1.3.2 Sediment Yield and Sediment Delivery

1.4 Clinton River Watershed Modeling System

1.4.1 Hydrologic Modeling

1.5 Channel Hydrodynamic and Sediment Transport Modeling

1.6 Summary

2.0 DATA COLLECTION

2.1 Introduction

2.2 Stakeholders Meeting Summary

2.3 Background and Previous Studies

2.4 Data Sources

2.4.1 Suspended Sediment Data

2.4.2 Precipitation Data

2.4.3 Discharge Data

2.5 GIS Data

3.0 SEDIMENT BUDGET

3.1 Introduction

3.2 Key Aspects of Sediment Delivery in the Clinton River Watershed

3.3 Flashiness Analysis

3.4 Rating Curve Analysis

3.5 Watershed Characterization System

3.6 Aerial Photography Evaluation

3.7 Summary

USACE – Detroit District Table of Contents Clinton River

4.0 HYDROLOGY AND SEDIMENT DELIVERY MODEL DEVELOPMENT ...48

4.1 Introduction

4.2 SWAT Model Description

4.2.1 SWAT Model Background

4.2.2 Characterization of Processes Using SWAT

4.2.3 ArcView SWAT Interface - AVSWAT

4.3 Clinton River Watershed SWAT Results

4.3.1 Paint and Galloway Subbasin Results

4.4 GSSHA Model Description

4.4.1 Precipitation

4.4.2 Precipitation Interception

4.4.3 Snowfall Accumulation

4.4.4 Infiltration

4.4.5 Evapotranspiration

4.4.6 Overland Flow Routing

4.4.7 Groundwater Flow

4.4.8 Overland Sediment Erosion / Deposition

4.4.9 Channel Routing

4.4.10 Instream Sediment Transport

4.5 Clinton River Subbasin Results With GSSHA

4.5.1 Paint Creek

4.5.2 Galloway Creek

4.5.3 Middle Branch Clinton River

5.0 ASSESSMENT OF BEST MANAGEMENT PRACTICES

5.1 Introduction to BMP Modeling

5.2 Model Setup

5.3 Existing Conditions

5.4 Effect of Lot Sizes

5.5 Effect of Small Scale Contributions

5.6 Effect of Controlled Runoff from Construction Sites

5.7 Effect of Buffer Strips

5.8 Effect of Retention Basins

5.9 Effect of Ditch Bank Erosion

5.10 Effect of Urbanization

5.11 Role of Wetlands

5.12 Effect of Sediment Traps

6.0 DETAILED INSTREAM SEDIMENT MODELING

6.1 EFDC Model Description

6.2 Model Setup

6.2.1 Model Domain and Grid Generation

6.2.2 Bed Sediment

6.2.3 Model Boundary Conditions

6.3 Model Calibration

6.3.1 Flow

6.3.2 Total Suspended Solids

6.3.3 Bed Change

6.4 Model Runs

6.4.1 Impact of Lake Seiche on Hydrodynamics

6.4.2 Impact of the Inflatable Weir

7.0 MODEL APPLICATION DISCUSSION

7.1 Model Characteristic Comparison

7.2 Comparison of SWAT and GSSHA Results

7.2.1 Discharge

7.2.2 Sediment Delivery

7.3 Recommendations for Model Application

8.0 CONCLUSIONS

8.1 Existing Conditions

8.2 Effects of Changing Land Use on Catchment Hydrograph Response and

Sediment Delivery

8.3 Impacts of Best Management Practices

8.4 Conclusions of Hydrologic Modeling Exercises

8.5 Impact of Lake Seiche and Inflatable Weir on River Hydrodynamics

9.0 REFERENCES

Appendix A - WCS Results

Appendix B - Process Characterization in the SWAT Model

Appendix C - CD "Visualization of GSSHA Results Using Spatial Data Analyzer (SDA)"

LIST OF FIGURES

Figure 1.1 Basin Topography

Figure 1.2 Relationship of Sediment Yield to Effective Precipitation (Langbein and Schumm, 1958; Knighton, 1998)

Figure 1.3 Relationship of Sediment Yield to Mean Annual Precipitation (Walling & Kleo, 1979)

Figure 1.4 Relationship Between Drainage Area and Sediment Delivery Ratio for Different Parts of the World (Walling, 1983; Knighton, 1998)

Figure 1.5 General Outline of the Clinton River Watershed Modeling System

Figure 1.6 Coverage of Model Domains in the Clinton River Watershed Modeling System

Figure 1.7 Clinton River Watershed Modeling System

Figure 2.1 Locations of Precipitation Gages

Figure 2.2 USGS Streamflow Gages

Figure 3.1 Distribution of Cropland in the Clinton River Watershed

Figure 3.2 Mean Annual Runoff for Four Different Land Use Types (Dunne, 1979)

Figure 3.3 Distribution of Developed Land in the Clinton River Watershed

Figure 3.4 Distribution of Forests and Wetlands in the Clinton River Watershed

Figure 3.5 Variation of Surface Slope in The Clinton River Watershed

Figure 3.6 Summary of Changes in River Flashiness

Figure 3.7a Gage # 04165500 Clinton River at Mt. Clemens

Figure 3.7b Gage # 04164000 Clinton River near Fraser

Figure 3.7c Gage # 04161540 Paint Creek at Rochester

Figure 3.7d Gage # 04161580 Stony Creek near Romeo

Figure 3.7e Gage # 04160800 Sashabaw Creek near Drayton Plains

Figure 3.7f Gage # 04161100 Galloway Creek near Auburn Hts

Figure 3.8 Relationship Between Suspended Sediment Load and Daily Discharge at USGS Gage # 04165500

Figure 3.9 Mean Suspended Sediment Discharge Estimated from Rating Curve and Discharge Record

Figure 3.10 WCS Soil Erosion Results for HUC14 Subbasins

Figure 3.11 WCS Sediment Yield and Delivery Results for HUC14 Subbasins

Figure 3.12 Relationship Between Percentage of Cropland in a Subbasin and Erosion Predicted by WCS

Figure 3.13 Relationship Between Percentage of Developed Land in a Subbasin and Erosion Predicted by WCS

Figure 3.14 Bank Erosion in Sterling Heights Park. Source: MDNR

Figure 3.15 1999-2000 Digital Orthophoto of Sterling Heights Park

Figure 4.1 Clinton River Subbasins Modeled in SWAT

Figure 4.2 SWAT Precipitation and Temperature Stations

Figure 4.3 Average Annual Flow Comparison of SWAT Results and USGS Measurements

Figure 4.4 Average Monthly Flow Comparison of SWAT Results and USGS Measurements

Figure 4.5 Average Daily Flow Comparison of SWAT Results and USGS Measurements

Figure 4.6 SWAT Model Daily Sediment Delivery Results for 1992

Figure 4.7 Clinton River Watershed 1978 Land Use Breakdown

Figure 4.8 Clinton River Watershed 1992 Land Use Breakdown

Figure 4.9 SWAT Yearly Net Erosion Per Unit Area Results for 1978 and 1992 Land Use Scenarios (with 1990’s climate data)

Figure 4.10 Average Daily Flow Comparison of SWAT Results for Paint Creek Watershed and USGS Measurements for 1996

Figure 4.11 Paint Creek Watershed Daily Sediment Delivery SWAT Results for 1996

Figure 4.12 Average Daily Flow Comparison of SWAT Results for Galloway Creek Watershed and USGS Measurements for 1989

Figure 4.13 Galloway Creek Watershed Daily Sediment Delivery SWAT Results for 1989

Figure 4.14 Paint Creek Subbasin Land Use, River Network and Gage Location

Figure 4.15 Precipitation and Flow Time Series for the June 18th, 1996 Event

Figure 4.16 Sediment Flux Time Series for the June 18th 1996 Event

Figure 4.17 Galloway Creek Subbasin Land Use, River Network and Gage Location

Figure 4.18 Results of Galloway Creek Calibration

Figure 4.19 Middle Branch Subbasin Land Use, River Network and Gage Location

Figure 4.20 GSSHA Model Calibration Results for USGS Gage #04164800

Figure 4.21 GSSHA Model Calibration Results for USGS Gage #04164600

Figure 5.1 Study Area Selected for BMP Assessment

Figure 5.2 TIN Map of the Study Area

Figure 5.3 Model Domain Comprised of 109 by 93 Cells in the X and Y Axis

Figure 5.4 Outlet Hydrograph and Sediment Load for Existing Scenario

Figure 5.5 Curve Numbers and % Impervious Area for Different Lot Sizes

Figure 5.6 Outlet Hydrograph and Sediment Load for Different Lot Sizes

Figure 5.7 Sediment Yield Over the Duration of a Storm for Different Lot Size Discharge Rates

Figure 5.8 ¼ Acre Lot Size Distribution and Land Use

Figure 5.9 Sediment Load Plots for Different Buffer Types and the Four Land Use Types

Figure 5.10 Sediment Load Plots for Different Land Use Types and the Five Buffer Types

Figure 5.11 Sediment Load Plots for Different Land Use Types and the Six Buffer Types

Figure 5.12 Discharge Plots for Short Grass Buffer Type with Different Land Use Types and the Three Buffer Widths

Figure 5.13 Topographic Contours for the Lower Reach of the Model Domain

Figure 5.14 Galloway Subbasin - Baseline (1992) Land Use Conditions

Figure 5.15 Galloway Subbasin - Modified Conditions

Figure 5.16 GSSHA Model Flow Results - Baseline vs. Urbanized

Figure 5.17 GSSHA Model Sediment Delivery Results - Baseline vs. Urbanized

Figure 6.1 Inflatable Weir Near to the Flow Diversion in the Lower Clinton River

Figure 6.2 Curvilinear Grid and Model Domain for Lower Clinton River

Figure 6.3 Interpolated Bed Elevation for Lower Clinton River

Figure 6.4a Fraction of Fine Gravel in the Lower Clinton River Bed Material

Figure 6.4b Fraction of Sand in the Lower Clinton River Bed Material

Figure 6.4c Fraction of Silt in the Lower Clinton River Bed Material

Figure 6.4d Fraction of Clay in the Lower Clinton River Bed Material

Figure 6.5 Bulk Density for the Lower Clinton River

Figure 6.6 Relationship Between Suspended Sediment Load and Daily Discharge at USGS Gage # 04165500 on Clinton River

Figure 6.7 Relationship between the Percentage of Sediment Finer Than 0.062mm and Daily Discharge at USGS Gage # 04165500 on Clinton River

Figure 6.8 Percentages of Clay, Silt and Sand Over Range of Daily Discharges at USGS Gage # 04165500 on Clinton River

Figure 6.9 Incoming TSS Concentrations for Sand, Silt and Clay Over Storm Event Run

Figure 6.10 USGS TSS, Velocity and Discharge Data From 1974-2002

Figure 6.11 Measured and Modeled Velocity vs. Discharge

Figure 6.12 Measured and Modeled TSS vs. Daily Discharge

Figure 6.13 a) Measured Bed Change (m) Between 1999-2001 (Above) and b) Modeled Bed Change (m) for One Storm Event (Below)

Figure 6.14 Lake St. Clair Lake Level (During Seiche)

Figure 6.15 Flow at Upstream Limit (Before Seiche)

Figure 6.16 Partial Flow Reversal at Upstream Limit (During Seiche)

Figure 6.17 Flow at River Mouth (Before Seiche)

Figure 6.18 Flow Reversal at River Mouth (During Seiche)

Figure 6.19 Distribution of Flow Over Canal Width

Figure 6.20 Operation State of Inflatable Weir

Figure 6.21 Bed Change With the Weir Inflated for a Storm Event on June 2000

Figure 6.22 Bed Change With the Weir Inflated for a Storm Event on July 2000

Figure 6.23 Bed Change With the Weir Deflated for a Storm Event on June 2000

Figure 6.24 Bed Change With the Weir Deflated for a Storm Event on July 2000

Figure 7.1 Daily Average Discharge Values for SWAT, GSSHA and USGS Gage

Figure 7.2 Sediment Delivery Results at Outlet From SWAT and GSSHA

LIST OF TABLES

Table 1.1 Examples of Natural Sources and Sinks of Sediment

Table 1.2 Examples of Anthropogenic Sources and Sinks of Sediment

Table 1.3 Summary of Sediment Yields from Other Studies

Table 1.4 Summary of Models in the Clinton River Watershed Modeling System

Table 1.5 Summary of CWMS Hydrological Models

Table 2.1 USGS Flow Gage Stations Within the Clinton River Watershed

Table 3.1 Trends in River Flashiness in the Clinton River Watershed

Table 4.1 SWAT Results for Net Erosion and Sediment Yield for Clinton River Watershed

Table 4.2 Hydraulic Conductivity Values Based on Soil and Land Use Type

Table 4.3 Manning’s Roughness Coefficient for Overland Flow

Table 4.4 Soil Erodibility Values

Table 4.5 GSSHA Model Infiltration Adjustment

Table 4.6 GSSHA Model Infiltration in cm/hour

Table 4.7 Manning’s Friction Coefficient for Overland Flow

Table 4.8 Middle Branch GSSHA Model Infiltration in cm/hour

Table 4.9 Manning’s Friction Coefficient for Overland Flow

Table 5.1 Friction, Infiltration and Erodibility Parameters for Various Land Uses and Loamy Sand Soil

Table 5.2 Ks and C Values for Different Lot Sizes Based on Curve Number and Percentage of Impervious Area

Table 5.3 Net Sediment and Water Volumes for Different Lot Sizes and Actual Conditions

Table 5.4 Net Sediment and Water Volumes for Rooftop Scenarios and Comparison with ¼ Acre Lots

Table 5.5 Erosion Control Structures for Construction Sites and Associated P Values and Erosion Volumes

Table 5.6a Infiltration for Different Buffer and Land Use Types

Table 5.6b Runoff Reduction for Different Buffer and Land Use Types

Table 5.6c Sand Erosion Reduction for Different Buffer and Land Use Types

Table 5.6d Sand Delivery Reduction for Different Buffer and Land Use Types

Table 5.6e Silt Erosion Reduction for Different Buffer and Land Use Types

Table 5.6f Silt Delivery Reduction for Different Buffer and Land Use Types

Table 5.6g Clay Erosion Reduction for Different Buffer and Land Use Types

Table 5.6h Clay Delivery Reduction for Different Buffer and Land Use Types

Table 5.6i Total Erosion Reduction (Sand/ Silt/ Clay) for Different Buffer and Land Use Types

Table 5.6j Total Delivery Reduction (Sand/ Silt/ Clay) for Different Buffer and Land Use Types

Table 5.7 Sedimentation Basin Dimensions, Trapping Efficiency and Flush Time to Achieve a 9.5 cm3 Outflow Load ............................................................................. 102 - 103

Table 5.8 Sediment and Water Volume Change for Different Ditch Bank Scenarios

Table 6.1 Maximum Cumulative Bed Change

Table 7.1 SWAT and GSSHA Attributes and Parameter Comparison

Table 7.2 SWAT and GSSHA Characteristics

LIST OF ACRONYMS

ADE Alternating Direction Explicit ADE-PC Alternating Direction Explicit with Prediction-Correction AOC Area of Concern ASCE American Society of Civil Engineers AVSWAT ArcView SWAT BASINS Better Assessment Science Integrating Point & Nonpoint Sources BMP Best Management Practice CN Curve Number CRWC Clinton River Watershed Council CWMS Clinton River Watershed Modeling System DEM Digital Elevation Model DEQ Department of Environmental Quality EFDC Environmental Fluid Dynamics Computer Code ELP Empirically based, Lumped Parameter models EPA Envrionmental Protection Agency ERDC Engineer Research and Development Center GA Green and Ampt GAR Green and Ampt with Redistribution GIS Geographic Information System GSSHA Gridded Surface-Subsurface Hydrologic Analysis HC Hydraulic Conductivity HD Hydrodynamic HRU Hydrologic Response Unit MLGA Multi-layer Green and Ampt MUSLE Modified Universal Soil Loss Equation NED National Elevation Dataset NLCD National Land Cover Dataset NOAA National Oceanic and Atmospheric Agency NPS Non-point Source NRCS Natural Resources Conservation Service NSI National Sediment Inventory NWS National Weather Service PCB Polychlorinated Biphenyl PDE Partial Differential Equation PDP Physically-based, Distributed Parameter Models PE Point Explicit RAP Remedial Action Plan SCS Soil Conservation Service SEMCOG Southeast Michigan Council of Governments SSL Suspended Sediment Load ST Sediment Transport STORET STOrage and RETrieval SWAT Soil Water Assment Tool SWRRB Simulator for Water Resources in Rural Basins TIN Triangulated Irregular Network TMDL Total Maximum Daily Load USACE United States Army Corps of Engineers USDA United States Department of Agriculture USDA-ARS United States Department of Agriculture - Agriculatural Research Station USEPA United States Environmental Protection Agency USGS United States Geologic Survey USLE Universal Soil Loss Equation WCS Watershed Characterization System WERF Water Environmental Research Foundation WY Water Year

EXECUTIVE SUMMARY

The U.S. Army Corps of Engineers is directed to develop sediment transport models for tributaries to the Great Lakes that discharge to Federal navigation channels or Areas of Concern (AOCs). These models are being developed to assist State and local resource agencies in evaluating alternatives for soil conservation and non-point source pollution prevention in the tributary watersheds. The ultimate goal is to support State and local measures that will reduce the loading of sediments to navigation channels and AOCs, and thereby reduce the costs for navigation maintenance and sediment remediation. This report includes a description of the Clinton River Watershed and the modeling tools that were developed and tested during the course of this study.

The Clinton River is located just north of Detroit in southeastern Michigan. The main channel traverses 80 miles (128 km) from its headwaters in the western portion of the watershed, to Lake St. Clair near the city of Mt. Clemens. The watershed covers 760 square miles (1,968 km2) of southeastern Michigan, including portions of Oakland and Macomb Counties and small areas of St. Clair and Lapeer Counties. The watershed is home to more than 1.6 million people in 56 municipalities. The southern portion of the watershed is dominantly urban, the middle section is undergoing rapid development of suburban land use, and the northern region is primarily agricultural and forested. The condition of the river and its tributaries varies dramatically, from runoff and pollution problems in urban areas, to healthier waters with thriving trout fisheries in rural areas.

The diversion canal in the downstream reach of the river is an important control on water and sediment movement in the lower watershed, and it also has a significant impact on sediment delivery to Lake St. Clair. The intent of the diversion canal was to alleviate flooding problems in Mt. Clemens. The diversion is controlled by an adjustable weir, which allows for the necessary flood conveyance in times of excess river flow, while forcing the flow down the original river channel during low flow. A detailed investigation of the impact of the canal and diversion structure on water and sediment transport is provided in Section 6 of this report.

There are no direct measurements of soil erosion or sediment delivery within the Clinton River Watershed. However, it is possible to estimate likely sediment yield and sediment delivery ratio for the Clinton River Watershed using basic relationships found in general watershed geomorphology literature. In addition, sediment yield estimates were generated using the Clinton River Watershed Model System (CWMS) developed during this study. There is a broad range in sediment yield estimates due to the large variation in methodologies used to estimate watershed soil erosion and channel sediment loads.

To assess potential management problems and to evaluate a wide range of best management practices (BMPs), a set of computational tools were used to study watershed hydrology, soil erosion, sediment delivery, river channel hydrodynamics and sediment transport. These models provide a general understanding of the hydrologic and geomorphic behavior of the watershed, allowing the prediction of the relative effects of

USACE – Detroit District 1 Clinton River changing land use and the effectiveness of different best management practice (BMP) strategies on subwatershed scale, soil erosion and sediment yield.

The three hydrologic models (WCS, SWAT, GSSHA) in the CWMS simulate overland flow generation, sediment detachment (soil erosion) and sediment yield from the watershed. They use input data from the Clinton River Watershed GIS and use a variety of modeling approaches to represent watershed processes. Each model has a different range of applicability and all three were necessary to characterize the different aspects of watershed processes examined in this study.

Land use and land use change are key factors governing soil erosion and sediment yield in this watershed. Cultivated and grazed land is shown to be the greatest non-point source of sediment, while developed land leads to flashy river flow. Urban and suburban development is likely to increase soil erosion (especially from construction sites) in the future, and this may be combined with greater riverbed and bank erosion from a flashier river regime to increase sediment yield. However, a reduction in the amount of sediment available for transport once an area is urbanized may serve to counteract this.

Urbanization affects soil erosion rates by potentially reducing the sediment available to be entrained from hillslopes, although this depends on many factors, such as the nature of the storm drainage system, degree of buffering, type of sewer outfalls, housing lot structure, etc. While process-response models suggest an inverse relationship between urbanization, soil erosion and sediment yield, these models ignore the mitigating effect of intrabasinal storage and/or sediment conveyance rates. In particular, increased flashiness of the urban hydrograph may be enough to trigger river channel instability and therefore increased channel erosion may balance reduced hillslope erosion.

Many watersheds have experienced high population growth rates during the post-war period, yet sediment yields are not strongly correlated to population or new housing development. Sediment yield, in general, is strongly controlled by the occurrence of peak flows. This implies that changes in sediment conveyance rates can mask changes in land use practices and vice-versa. In a broader context, this underscores the need to address anthropogenic impacts on watershed sediment yield when considering BMP actions.

Obviously, future sediment yield is an important component in land use planning, and to state simply that urbanization reduces sediment yield would undermine the need for careful consideration of process interactions as urban development takes place.

Increased sedimentation rates may also be expected in the relatively flat, lower section of the watershed as development continues. These factors serve to stress the importance of appropriate choices of BMPs in this watershed and highlight the utility of using a numerical modeling approach for evaluating their implementation.

Direct assessment of sediment yield and sediment delivery from empirical evidence was not possible in the Clinton River Watershed. Comparison of catchment baseline conditions (such as drainage area; effective precipitation etc.) with empirical

USACE – Detroit District 2 Clinton River relationships established by other studies give estimates of sediment yield in the order of 200-600 t/km2 yr-1.

The SWAT model predicts average net erosion for the years 1990-1994 of 380 t/km2 yr-1.

This value seems reasonable, compared to the empirical estimates above. Mean annual sediment delivery at the SWAT model outlet (near gage #04165500) for 1990-1994 was 539,000 t/yr, giving an estimated sediment yield of 270 t/km2 yr-1.

Estimates of watershed sediment load using rating curves for the most downstream station in other Great Lakes watersheds have suggested sediment yields of 15-100 t/km2 yr-1. Mean daily suspended sediment load at USGS Gage #04165500 is 82.8 t/day, suggesting a mean annual suspended sediment delivery in the region of 30,000 t/yr. This leads to an estimate of mean annual suspended sediment yield of 16 t/km2 yr-1 at this point in the watershed. The main cause of lack of agreement between sediment yields derived from empirical relationships, model predictions and rating curve analysis is unclear at present. This ambiguity reflects the lack of sediment load data within the watershed.

Substantial portions of the watershed have undergone a shift from primarily agricultural to urban land use. A flashiness index was computed for all stream gage records in the Clinton River Watershed that had a minimum of 20 years of data. The results indicate that several of the gages in the southern portion of the watershed have shown marked increases in flashiness since 1970. Other gages have remained steady or show minor decreases in flashiness with time. As expected, this suggests that there is a close link between the development of land in the southern part of the watershed and an increase in flashiness. The reduction in flashiness elsewhere in the catchment may be explained by a shift from agricultural to forestland use causing a decline in overland flow, through flow and therefore rapid hydrograph response.

There are several processes by which increased flashiness in hydrograph response may increase sediment loads in the watershed. Increased flashiness suggests increased overland flow, increasing erosion rates and increasing sediment transport from hillslope to channel. This may be balanced by a shift from agricultural to urban land uses, which may reduce sediment supply at the hillslope scale. Urban runoff may contain significant amounts of sediment, pollutants and nutrients if stormflow through sewers is left untreated before it discharges into the river channel. In addition, developed land has often meant the eradication of buffer strips by riparian landowners, which may cause a significant increase in the amount of sediment reaching a river channel.

A wide variety of BMP alternatives were evaluated using the CWMS. The large-scale effects of changing land use over time on watershed sediment yield and sediment delivery were investigated using the SWAT model. Analysis of the relative effectiveness of riparian buffer strips was the most extensive of the BMP analyses. Different buffer widths and vegetative types were evaluated for different land use types surrounding the buffer zone using the GSSHA model. This analysis identified the most effective vegetative types for buffer implementations as well as sensitivity to the different

USACE – Detroit District 3 Clinton River vegetation types and buffer widths. Variations in vegetation cover did not produce as much variation in sediment delivery as changes to buffer width.

Change in urban density was evaluated by changing lot sizes within the GSSHA hydrologic model. GSSHA predicted that with increasing urban density, runoff plays a more important role in erosion than the available sediment for erodibility. In other words, although the erodible soil would be reduced with urbanization (lawns and driveways) the increased runoff provides a more erosive mechanism for sediment transport.

In an ideal situation there would be sufficient sediment and flow data to calibrate the models. It is usually the case that these data are very limited, thus a more thorough quantitative calibration of the model is of questionable merit until further sediment load data are available. However, the lack of final calibration of the model system does not necessarily diminish the usefulness of the system in BMP planning as the system still allows for comparison of different management options and scenarios.

The different numerical models implemented for this watershed have different functional advantages and disadvantages; thus the generation of a watershed modeling system provides the advantage of a versatile and integrated tool to aid in the management of the basin. It is recommended that this set of tools is applied with due diligence since the interpretation of model results requires experience in the fields of hydrology, geomorphology and sediment dynamics.

SWAT is a very useful tool for long-term, large-scale watershed modeling. The input data for SWAT are readily available and the model can be calibrated to measured data with a reasonable amount of effort. The ability of SWAT to simulate nutrients and contaminants makes it a valuable tool for comprehensive watershed management and BMP evaluation.

The application of GSSHA at a subwatershed scale provided results reasonably similar to measured flow data. The effort required for this analysis is relatively high and should be applied for short-term simulations such as storm events or hypothetical situations to evaluate land use changes. The application of GSSHA to assess wetlands and sediment traps was not effective with the current version. However, a new edition of GSSHA, which has the built-in capability to account for wetlands, lakes and reservoirs, is released as a beta version.

The three-dimensional channel hydrodynamics and sediment transport model EFDC was useful in assessing hydrodynamics and sediment transport in the Lower Clinton River.

However, it is a computationally intensive tool and should be applied for very specific purposes such as assessing the impact of the inflatable weir on channel sedimentation.

Increased urbanization and flashiness of the river system (especially without adequate BMPs to compensate) may cause further challenges to the management of water and sediment in the Lower Clinton River. This is in turn related to management of the

USACE – Detroit District 4 Clinton River inflatable weir. EFDC could play an important role in informing the future active control of the weir to address related sedimentation and flooding issues.

USACE – Detroit District 5 Clinton River

This page is intentionally left blank.

1.0 INTRODUCTION

1.1 Purpose

The U.S. Army Corps of Engineers is directed to develop sediment transport models for tributaries to the Great Lakes that discharge to Federal navigation channels or Areas of Concern (AOCs). These models are being developed to assist State and local resource agencies in evaluating alternatives for soil conservation and non-point source pollution prevention in the tributary watersheds. The ultimate goal is to support State and local measures that will reduce the loading of sediments to navigation channels and AOCs, and thereby reduce the costs for navigation maintenance and sediment remediation. This report includes a description of the Clinton River Watershed and the modeling tools that were developed and tested during the course of this study.

1.2 Watershed Description

The Clinton River is located just north of Detroit in southeastern Michigan. The main channel traverses 80 miles (128 km) from its headwaters in the western portion of the watershed, to Lake St. Clair near the city of Mt. Clemens. The watershed covers 760 square miles (1,968 km2) of southeastern Michigan, including portions of Oakland and Macomb Counties and small areas of St. Clair and Lapeer Counties. The basin has two distinct topographic regions, the upper half, which has more relief and steeper channels, and the lower half, characterized by flatter topography and channel slopes. Although the entire basin covers parts of four counties, the two regions coincidentally occupy different political boundaries as well. The upper region lies predominantly in Oakland and Lapeer Counties, and the lower region in Macomb and St. Clair Counties (Figure 1.1).

The watershed is home to more than 1.6 million people in 56 municipalities. The urban development status within the watershed is varied: the southern portion is urban, the middle section is made up of rapidly developing suburbs and the northern region is rural.

The condition of the river and its tributaries varies dramatically, from runoff and pollution problems in urban areas, to healthier waters with thriving trout fisheries in rural areas. Sediment, stormwater quantity, and bacteria continue to be the most prominent pollutants challenging water quality and habitat.

USACE – Detroit District 6 Clinton River

Figure 1.1 Basin Topography

The diversion canal in the downstream reach of the river is an important control on water and sediment movement in the lower watershed, and it also has a significant impact on sediment delivery to Lake St. Clair. The intent of the diversion canal was to alleviate flooding problems in Mt. Clemens. While the diversion canal provided flooding relief, it has required modification in more recent years to facilitate active control over the diversion. The control is provided by an adjustable weir, which allows for the necessary conveyance in times of excess river flow, while forcing the flow down the original river channel during low flow. The structure (as originally built) diverted too much of the low flow down the bypass channel, which resulted in unacceptably low flow rates down the original channel during dry weather. The same phenomenon was also partially responsible for the buildup of sediments upstream of the diversion structure. A detailed investigation of the impact of the canal and diversion structure on water and sediment transport is provided in Section 6 of this report.

1.3 Clinton River Watershed Sediment Delivery

Sediment in a river can be transported by two processes: suspension and bedload transport. Suspended sediment load in the watershed is usually dominated by clay- and silt-sized particles. Typical natural processes involved in the erosion and delivery of suspended sediment are:

USACE – Detroit District 7 Clinton River

Overland flow;

Sheet and rill erosion;

Gully erosion;

Riverbed erosion;

River bank erosion, particularly shearing of fine material by river flow, and bank collapse.

These processes are generally distinguished into channel and non-channel sources and also into point and non-point sources. They may be accelerated by human activity in the catchment, as agriculture, industry and construction tend to mobilize and supply additional sediment to the watershed. In agricultural regions, soil loss from cultivated areas may dominate watershed suspended sediment delivery.

The relative contributions of the above different processes vary with location in the watershed. Headwaters of the catchment have steeper slopes without floodplains, and these areas are strongly coupled to channels (i.e. there are few sediment sinks between the source areas and the channel). Non-point sources therefore tend to dominate these areas and sediment is largely from non-channel sources. This may have implications for other watershed management issues, such as non-point source contaminant movement in upper watershed zones. Further downstream, wider floodplains increase temporary sediment storage and the channel becomes increasingly decoupled from the surrounding hillslopes. In these areas, in-channel sediment sources dominate.

Various sediment sinks also exist in the catchment. These may be either natural or anthropogenic, and they are often classified into temporary or permanent sinks. An example of the former would be channel storage of sediment in bar deposits, as sediment may be stored for a period of time before being remobilized by a high magnitude event.

An example of a permanent sink would be the abstraction of sand or gravel from the channel for use as construction material. Natural sediment sources and sinks are listed in Table 1.1, and anthropogenic sources and sinks are listed in Table 1.2. The natural processes listed in Table 1.1 are likely to be heavily modified by human activity, which may lead to acceleration or reduction of erosion and delivery rates, depending on the type and extent of the modification.

1.3.1 Terminology

The terms ‘sediment yield’ and ‘sediment delivery’ have often been used synonymously, while in other instances have been used to describe different aspects of watershed sediment movement. This section therefore provides definitions of these (and related) terms in order to be consistent throughout this report.

Sediment production (a.k.a. total denudation) is the amount of material made available for transport by weathering processes in a catchment. Note that weathering is an in-situ process and does not imply a movement of sediment through the watershed.

USACE – Detroit District 8 Clinton River

Net erosion is the amount of material mobilized in the catchment by hillslope erosion processes such as sheet and rill erosion. It is synonymous with ‘soil erosion’ in this report. While wind erosion can be a significant geomorphic agent in some areas, it was not included in any numerical calculations in this report, and is excluded from soil erosion estimates. Erosion typically has the units of tons/year. Soil erosion may also be expressed as a specific value (i.e. total amount divided by area of measurement) in tons/square mile/year.

Sediment yield is the amount of sediment passing a specified channel location and is influenced by a number of geomorphic processes. It may be substantially less than the amount actually eroded in the basin. Sediment yield is expressed as the total sediment volume delivered to a specified location in the basin, divided by the effective drainage area above that location for a specified period of time. Yield typically has the units of cubic meters/square kilometer/year or tons/square mile/year. Occasionally it is also necessary to estimate yield from a watershed from individual storm events of specified frequency. In some watersheds, single event sediment yields often exceed average annual values by several orders of magnitude.

Sediment delivery is the amount of sediment per year reaching the watershed outlet. It usually has the units of cubic meters/year or tons/year.

Sediment delivery ratio is the sediment yield divided by the gross amount of erosion occurring in the watershed upstream. This typically has values of less than 1.

1.3.2 Sediment Yield and Sediment Delivery

There are no direct measurements of net erosion, sediment yield and sediment delivery within the Clinton River Watershed. However, it is possible to estimate likely sediment yield and sediment delivery ratios for the Clinton River Watershed using basic relationships found in general watershed geomorphology literature. These estimates are intended as background information for the study due to the lack of sediment transport data, and the consequential lack of model calibration data.

USACE – Detroit District 9 Clinton River

Table 1.1 Examples of Natural Sources and Sinks of Sediment Sources Sinks Upland sheet and rill erosion Colluvial deposition Upland gully erosion Redistribution on upland valley slopes Channel bank erosion (several mechanisms) Floodplain storage Channel bed scour Channel storage Animal action (e.g. burrowing) Lakes Landslides

Table 1.2 Examples of Anthropogenic Sources and Sinks of Sediment Sources Sinks Cultivated land Dams and reservoirs Managed Forests Sand and gravel extraction Forest roads Hedgerow planting Urban areas – including drainage systems Vegetation management Removal of vegetation for construction Conservation activities Roads and road drainage systems Agricultural management (e.g.

agroforestry, Construction sites contour cultivation, etc.)

Mining Mining – Indirect (e.g. land devegetated by smelter fumes Spoil Heaps Vegetation fires Channel modification Mismanagement Indirect effects (e.g. devegetation through climate change)

Suspended sediment is normally the dominant component of sediment yield, with the exception of very extreme storm events. Langbein and Schumm (1958) developed a general relationship between effective precipitation and sediment yield for group-averaged data from the United States (Figure 1.2). Effective precipitation is that part of the total hydrograph which constitutes ‘quickflow’, as in the definition of the Unit Hydrograph (see Dunne & Leopold, 1978 for a review). Effective precipitation is also called ‘quickflow’, or ‘precipitation excess’, and is a notion of Hortonian Overland Flow:

i.e. it is the proportion of precipitation that is ‘effective’ in producing rapid response streamflow (stormflow).

USACE – Detroit District 10 Clinton River

Figure 1.2 Relationship of Sediment Yield to Effective Precipitation (Langbein and Schumm, 1958; Knighton, 1998)

Macomb and Oakland Counties have a mean annual precipitation of 31.5 inches (800mm). 60-70% of this was assumed to be effective precipitation. This estimation was loosely on the SCS curve method (USDA-SCS, 1972). The SCS method requires information concerning the cropping practice, soil characteristics, pre-rainfall moisture status, and the rainfall amount. A visual appraisal of SCS curves suggested that a value of 60-70% was reasonable for this catchment as a first-order approximation. Hydrograph separation was outside the scope of this study and would not have improved confidence in the estimates in this section, given the high degree of approximation involved. The effective precipitation was therefore estimated to be in the range of 480-560 mm, giving an estimated yield of around 200-550 t km-2 yr-1. There is a broad range in this estimate due to the large difference in sediment load estimates, depending on whether reservoir or river sediment load data are used to evaluate the relationship. This difference is methodological, and may depend on many issues (such as how reservoirs trap sediment, underestimating channel transport, etc.).

For the relationship established by Walling and Kleo (1979; Figure 1.3), the estimated sediment yield for the Clinton River Watershed would be approximately 200 t km-2 yr-1.

Langbein and Schumm’s relationship was established for semi-arid regions in the American southwest, whereas Walling and Kleo reviewed the validity of the Langbein relationship for a broader range of climatic regimes. While the two relationships are similar in the lower range of precipitation values, Walling and Kleo suggest there is another peak in yield at 1500 mm mean annual precipitation. This is due to vegetation dynamics changing with climate (1300-1500 mm precipitation = Mediterranean and 2500mm = tropical monsoon). Walling and Kleo’s analysis may apply more to climates with strong seasonality, and Langbein and Schumm’s may be more continental, but both give similar predictions of yield for the Clinton River Watershed.

USACE – Detroit District 11 Clinton River

Figure 1.3 Relationship of Sediment Yield to Mean Annual Precipitation (Walling and Kleo, 1979)

The yields suggested by empirical relationships are markedly higher than the yields generated by analysis of sediment rating curves and river gage data from other Great Lakes watersheds (Table 1.3). That figures may in part differ from the large number of dams in these systems, which will reduce yields significantly below natural basins of similar size.

Table 1.3 Summary of Sediment Yields from Other Studies Watershed Drainage Area (km2) Mean Annual Sediment

Yield (t km-2 yr-1) Saginaw River (MI) 22,360 20

St. Joseph River (MI/IN) 12,400 15

Sandusky River (OH) 3,250 75

Black River (OH) 1,200 100

Nemadji River (MN/WI) 1,125 25

Menomonee River (WI) 350 40

The gross sediment yield from a basin will generally represent a small amount of the net erosion within a catchment, since much of the eroded material within a catchment will be deposited before reaching the outlet. The sediment delivery ratio is dependent on a wide range of natural and anthropogenic factors, including:

Climate;

USACE – Detroit District 12 Clinton River

Drainage area;

Basin topography;

Lithologic structure;

Vegetation cover;

Land use and land management.

The relationship between drainage area and sediment delivery ratio for several areas of the world is shown in Figure 1.4. Estimating from relationship six in Figure 1.4, with a drainage area of approximately 2,000 km2, the expected sediment delivery ratio for the Clinton River Watershed would be approximately 10%.

Figure 1.4 Relationship Between Drainage Area and Sediment Delivery Ratio for Different Parts of the World (Walling, 1983; Knighton, 1998)

1.4 Clinton River Watershed Modeling System

To assess the problems outlined above, a set of computational tools were used to evaluate watershed hydrology, net erosion, sediment delivery, river channel hydrodynamics and sediment transport. These models were developed to gain a general understanding of the hydrologic and geomorphic behavior of the watershed and to predict the effects of changing land use and the effectiveness of different best management practice (BMP) strategies on subwatershed scale erosion and sediment delivery. Models were calibrated against river flow and sediment transport records, reservoir sedimentation surveys and harbor dredging records. A summary of the models contained in the Clinton River Watershed Modeling System (CWMS) is presented in Table 1.3. A flow chart outlining the functionality of the CWMS is shown in Figure 1.5, and a map of model domain coverage is shown in Figure 1.6. The range of applicability of each model, the types of management issues each model addresses and the scale and complexity of the processes they represent are summarized in Figure 1.7.

USACE – Detroit District 13 Clinton River

Table 1.4 Summary of Models in the Clinton River Watershed Modeling System

Acronym Model name Agency Processes SWAT Soil Water Assessment Tool USDA / EPA Hydrology

Soil Erosion Sediment Delivery

WCS Watershed Characterization System

EPA Hydrology Sediment Yield Sediment Delivery

GSSHA Gridded Surface-Subsurface Hydrologic Analysis

USACE Hydrology 1D River Hydrodynamics and Sediment Transport Sediment Yield Sediment Delivery

EFDC Environmental Fluid Dynamics Computer Code

USEPA 3D River Hydrodynamics Sediment Erosion, Transport and Deposition

USACE – Detroit District 14 Clinton River

Data Manager (GIS; Data

Visualization)

Hydrologic Model

Rainfall

River HD/ST Model

Estuary HD/ST Model

Watershed Character.

Runoff;

Sediment delivery

Flooding, Sed. Budget

Storm Surge

Model output

Met.

Data

Satellite Data

River Data

Receiving Water data

Bathymetry

River Network

Cross Sections

EFDC

WCS, SWAT, GSSHA

Data Manager (GIS; Data

Visualization)

Hydrologic Model

Rainfall

River HD/ST Model

Estuary HD/ST Model

Watershed Character.

Runoff;

Sediment delivery

Flooding, Sed. Budget

Storm Surge

Model output

Met.

Data

Satellite Data

River Data

Receiving Water data

Bathymetry

River Network

Cross Sections

EFDC

WCS, SWAT

GSSHA

Figure 1.5 General Outline of the Clinton River Watershed Modeling System

Figure 1.6 Coverage of Model Domains in the Clinton River Watershed Modeling System

USACE – Detroit District 15 Clinton River

M O

D E

L A

N D

R A

N G

E

F

A

PP

L

IC

A B

IL

IT

Y

D E

L S

C A

L E

S A

N D

C

M

PL

E

X

IT

Y W

A T

E R

SH

E

D

IS

SU

E

S

E

FD

C

W C

S

R E

SO

L

U T

IO

N SC

A L

EC

O

M

PL

E X

IT

Y PA

R A

M E

T

E R

S

SP

A

T

IA

L C

O V

E R

A G

E

T E

M

PO

R A

L C

O V

E R

A G

E

R

IV

E R

C H

A N

N

L

A

N A

G E

M E

N T

R

IV

E R

E R

O

SI

O

SE

D

IM

E

N T

A T

IO

N

D E

T A

IL

E

D B

M Ps

A G

R

IC

U L

T U

R A

L L

A N

D M

A N

A G

E M

E

T

N D

U

SE

C H

A N

D

P L

A N

N

IN

IM

A

T E

C H

A N

G E

G

SS

H A

SW

A

T

Fi gu re

.7 C lin to n Ri ve r W at er sh ed

M od el in g Sy ste m

U

SA

C E

D et ro it D is tri ct lin to n R iv er re at

L ak es H yd ra ul ic s a nd

H yd ro lo gy O ff ic e Se di m en t T ra ns po rt

M od el in g St ud y

1.4.1 Hydrologic Modeling

The three hydrologic models (WCS, SWAT, GSSHA) in the CWMS simulate overland flow generation, soil erosion and sediment delivery from the watershed. They use input data from the Clinton River Watershed GIS and use a variety of modeling approaches to represent watershed processes (Table 1.5). Each model has a different range of applicability and all three were necessary to characterize different aspects of watershed processes examined in this study.

Table 1.5 Summary of CWMS Hydrological Models

Model Developer Hydrology Soil Erosion

Sediment Delivery

Spatial Resolution

Applicability

GSSHA USACE PDP /

Multi-dimensions

USLE

based

Yang (1973) method

Varies details / small catchments

Event / detail plan / detail design

SWAT USDA/EPA ELP / PDP MUSLE Process / Channel Sediment Routing

Medium / large watershed

Continuous / overview / agriculture practices / climate change / land use

WCS EPA ELP USLE Distance-and area-based equations

Large watershed

Annual based / quick overview

(ELP - Empirically based, lumped parameter models; PDP - Physically-based, distributed parameter models; USLE – Universal soil loss equation; MUSLE – Modified USLE)

The Watershed Characterization System (WCS) is an Arcview-based system designed to provide tools and an initial set of watershed data for characterizing and thereby understanding a watershed. It can be used to assist users in completing the watershed characterization phase required in developing Total Maximum Daily Loads (TMDLs).

This may include:

Characterization of the physical and hydrologic properties of the watershed, such as soil, land use, elevation, climate, and streamflow.

Evaluation of ambient water quality conditions, including inventory of monitoring stations and statistical analysis of observed data.

USACE – Detroit District 17 Clinton River

Assessment of potential sources of impairment, such as permitted dischargers, crop and livestock agriculture, mining, silviculture, and populated places, and preliminary estimation of pollutant loads from these sources.

WCS provides a simple and rapid method for estimating soil erosion and catchment sediment delivery on an average annual basis. For the Clinton River Watershed, results were determined for 41 subbasins within the Clinton River Watershed, based on 1992 land use data. When WCS is used for its intended purpose, in predicting long-term erosion and sediment delivery rates, it performs satisfactorily since the model is based on annual USLE predictions and basic empirical sediment delivery relationships. However, it should not be used on shorter timescales or for event-based predictions. The coarse nature of the WCS grid (30 m) also makes it unsuitable for detailed examinations of best management practices. Section 3.3 contains more detail on the WCS model.

The WCS model was valuable in carrying out a rapid, initial assessment of watershed sediment movement. However, it is limited in that it only predicts average annual sediment delivery rates, which may be predicted in a more flexible manner using SWAT.

WCS is therefore not included in the Clinton River Watershed Modeling System that accompanies this report.

The objective of SWAT is to predict the impact of management decisions on water and sediment yields in large complex watersheds. Variation in land use and management conditions over long periods of time can be represented. While the project scope did not include a whole-watershed model using SWAT, this was included instead of the 1D HEC-6 river modeling, which was originally scoped. SWAT is particularly good at modeling NPS-pollution loads from agricultural practices, and routines for urban sediment loads are currently being improved.