Development of historic and synthesized unregulated streamflow for the James River in North Dakota and South Dakota, 1983-91

Abstract

Operation of the Garrison Diversion Unit may have some affect on the hydrology of the James River in North Dakota and South Dakota. The Garrison Diversion Unit Monthly Operations Model was developed to analyze a wide range of streamflow conditions that could occur in the James River Basin. The purpose of this study was to compute monthly streamflows that are required as input to the model. Historic streamflow data were complied and record extension methods were used, when necessary, to compute monthly streamflow for 1983-91 for 15 gaging stations on the James River in North Dakota and South Dakota. The record extension methods used include Maintenance of Variance Extension Type 1, Ordinary Least Squares, and drainage-area ratio. In addition to the historic streamflow, synthesized unregulated streamflow was computed for the 15 gaging stations on the James River for 1983-91 by eliminating the effects of Jamestown Reservoir, Pipestem Reservoir, Sand Lake National Wildlife Refuge, consumptive surface-water withdrawals, and wastewater withdrawals. Maintenance of Variance Extension Type 1, Ordinary Least Squares regression, water-balance procedures, and drainage-area ratio method were used to compute the unregulated streamflows. INTRODUCTION In September 1983, a technical team consisting of members from the U.S. Bureau of Reclamation, U.S. Fish and Wildlife Service, U.S. Geological Survey, North Dakota State Water Commission, South Dakota Department of Water and Natural Resources, and Garrison Conservancy District met to discuss issues related to the hydrology of the James River Basin in North Dakota and South Dakota. The discussion focused on how to best analyze the effects that operation of the Garrison Diversion Unit might have on the hydrology of the James River in North Dakota and South Dakota. The Garrison Diversion Unit in North Dakota was authorized by Congress in 1965 to provide for irrigation, municipal and industrial water, fish and wildlife, recreation, and flood control. Although earlier studies to analyze the effects of the Garrison Diversion Unit on the hydrology of the James River have been completed by the U.S. Bureau of Reclamation, results obtained by using the Hydrologic River Operation Study System differed from those obtained by using the Return Flow Models (U.S. Department of the Interior, 1983, page HI-20). Because of the limitations of existing streamflow models, the technical team concluded that a new model was needed to analyze a wide range of streamflow conditions and the resulting hydrologic conditions that could be expected to occur in the James River Basin. The model, Garrison Diversion Unit Monthly Operations Model, developed by the U.S. Bureau of Reclamation required, as input data, the monthly unregulated stream flows at 13 gaging stations on the James River for 1953-82. Results of a study conducted to compile and analyze the 1953-82 monthly stream flow data needed as input to the model were published by Wiche, Benson, and Emerson (1989). Since the model was developed, two additional gaging stations have been installed on the James River, and an additional 9 years of stream flow data are available for many of the stations. Consequently, the U.S. Bureau of Reclamation needed an updated data set for the model. Stream flow data of different record lengths are available for the 15 gaging stations along the James River in North Dakota and South Dakota. This report provides the results of a study to compile and estimate the 1983-91 monthly stream flow data needed as input to the Garrison Diversion Unit Monthly Operations Model developed by the U.S. Bureau of Reclamation. The specific objectives of this study were as follow: (1) Compile and extend the streamflow record where necessary to develop 1983-91 monthly stream flow for the following 15 gaging stations on the James River (pi. 1, in pocket) Station number Station name 06468170 James River near Grace City, N. Dak.; 06468250 James River above Arrowwood Lake near Kensal, N. Dak.; 06468500 James River near Pingree, N. Dak.; 06470000 James River at Jamestown, N. Dak.; 06470500 James River at LaMoure, N. Dak.; 06470878 James River at North Dakota-South Dakota State line; 06471000 James River at Columbia, S. Dak.; 06472000 James River near Stratford, S. Dak.; 06473000 James River at Ashton, S. Dak.; 06475000 James River near Redfield, S. Dak.; 06476000 James River at Huron, S. Dak.; 06477000 James River near Forestburg, S. Dak.; 06478000 James River near Mitchell, S. Dak.; 06478500 James River near Scotland, S. Dak.; 06478513 James River near Yankton, S. Dak.; and (2) estimate 1983-91 monthly unregulated stream flows by using statistical or water-balance methods for all 15 gaging stations identified in item 1. In this study, an attempt was made to provide reasonable estimates of streamflow in terms of means and variances of the flow. However, the methods employed do not attempt to preserve the interstation correlation of streamflow values on the James River. DESCRIPTION OF STUDY AREA The James River, about 747 miles in total length, drains parts of east-central North Dakota and South Dakota (pi. 1). The James River Basin encompasses about 22,000 square miles, of which 8,000 square miles is in North Dakota and 14,000 square miles is in South Dakota. The headwaters of the James River are located in Wells County, N. Dak. From these headwaters, the James River extends about 100 miles to Arrowwood National Wildlife Refuge. Streamflow is low or nonexistent much of the year from the headwaters to the refuge. Jamestown Reservoir, located in a narrow valley immediately downstream of the refuge, was created by the completion of Jamestown Dam in 1953. Pipestem Creek, a major tributary to the James River, enters the James River 1 mile downstream of Jamestown Dam. Pipestem Creek has been regulated since the completion of Pipestem Dam and Reservoir in 1974. Downstream from Jamestown Dam, the James River meanders within the confines of a broad valley for about 135 miles to near the North Dakota-South Dakota State line. Near the State line, the James River enters the lakebed of glacial Lake Dakota and flows through Dakota Lake National Wildlife Refuge and Sand Lake National Wildlife Refuge. Both refuges were created by construction of low-head dams. The water-surface elevation in Dakota Lake National Wildlife Refuge is regulated by a low-head concrete dam control and the total storage capacity is about 3,200 acre-feet. The water-surface elevation in Sand Lake National Wildlife Refuge is regulated by two control structures, and the total storage capacity is about 27,200 acre-feet. After leaving Sand Lake National Wildlife Refuge, the river flows through the Lake Dakota Plain for about 150 miles to Redfield, S. Dak. Within parts of the Lake Dakota Plain, the slope of the river is less than 0.1 foot per mile and the channel capacity is as little as 200 cubic feet per second. Downstream from Redfield, S. Dak., the slope and channel capacity increase. The elevation of the river decreases about 130 feet in 474 river miles within South Dakota. Several other dams in addition to the low-head dams are located on the James River in South Dakota. The Tacoma Park and Spink County Dams are used primarily for recreation. 1\vo dams near Huron, the James Diversion Dam and the Third Street Dam provide the city of Huron's major water supply. Most of the smaller dams were constructed privately and are used as river crossings or as diversion points for private irrigation. METHODS OF STREAMFLOW RECORD EXTENSION Record extension of measured stream flow or synthesized unregulated stream flow can be accomplished by several methods, including use of drainage-area ratio, regional statistics, regression, and precipitation-runoff modeling. The drainage-area ratio method (Hirsch, 1979) is based on the assumption that the ratio of the flows at two sites is equal to the ratio of their drainage areas by using the following equation: where y,is the estimated flow during month t at the site of interest, in cubic feet per second; ay is the drainage area at the site of interest, in square miles; ax is the drainage area at the base station, in square miles; and xi is the gaged flow during month / at the base station, in cubic feet per second. If there are no records of streamflow at the site of interest or no reliable regional flow statistics, the assumption that the flow is proportional to the drainage area may be the best assumption to make. The regional-statistics method (Thomas and Benson, 1970) uses regional regression equations developed between streamflow and basin characteristics to estimate mean monthly flows. The equations have not been developed for North Dakota and South Dakota, and their development was not within the scope of this study. Therefore, this method was not used. Regression methods can be used to extend streamflow records in time. Regression methods are based on the assumption that streamflow records are available at the site of interest for a period of NI years, and records at the base station are available for the same N years, plus an additional A^ years. The use of untransformed data or logarithms to extend records was studied by Hirsch (1979) and Stedinger (1980). The conclusion reached by both Hirsch and Stedinger was to work with logarithms. Ordinary Least Squares (OLS) regression can be used to estimate stream flow at the site of interest by using the following equation: yi = a + bxi (2) where y t and xl are defined before. The parameters a and b are those values that minimize the squared errors. The solution of equation 2 becomes