Stream-Gaging Program of the U.S. Geological Survey
Reston, Virginia, 1995
By Kenneth L. Wahl, Wilbert O. Thomas, Jr., and Robert M. Hirsch


The stream-gaging program of the USGS does not represent a single "network" of stations, but is an aggregation of networks and individual streamflow stations that originally were established for various purposes. Because the data from about 4,200 of the 7,292 stations are telemetered by an earth-satellite-based communications system, those data are available in realtime for many agencies to conduct water-resources projects and for the National Weather Service (NWS) to forecast floods. Data from the active stations, as well as from discontinued stations, are stored in a computer data base that currently holds mean daily-discharge data for about 18,500 locations and more than 400,000 station-years of record, or more than 146 million individual mean daily-discharge values. Additional data are added to the data base each year. The stream-discharge data base is an ever-growing resource for water-resources planning and design, hydrologic research, and operation of water-resources projects. Increasing the length of individual station records is valuable for at least two reasons. Additional years of record provide ever-improving accuracy of estimates of streamflow characteristics, such as the magnitude of extreme infrequent floods or low flows, and an opportunity to determine how streamflow characteristics are changing over time due to such causes as agricultural practices, urbanization, ground-water development, or climate change.

Figure 2.
Number of stations operated by the U.S. Geological Survey in 1994, by State or possession.

Funding the Program

Just as the network of stations represents an aggregation, so does the program funding. Operating funds for individual stations in the program may come from a blend of Federal funds appropriated to the USGS, funds from State and local agencies, and funds appropriated to other Federal agencies ( Condes, 1994 ). Federal funds used for hydrologic data-collection activities of the USGS come from the following primary sources: funds made available by Congress to the USGS for matching State or local agency offerings under the USGS Federal--State Cooperative Program (herein referred to as the "Cooperative Program"), transfer of funds from other Federal agencies to meet their water-resources-data needs, and funds appropriated by Congress and designated specifically for use by the USGS for collection of streamflow and water-quality data.

More than 50 percent of the 7,292 stations operated by the USGS are funded through the Cooperative Program (fig. 3). Under that program, the USGS provides up to 50 percent of the funds, and the State or local agency provides the remainder. Currently, more than 600 State and local agencies participate in the stream-gaging program. Other stations in the program are operated by the USGS and funded by other Federal agencies, such as the U.S. Army Corps of Engineers (COE) and the Bureau of Reclamation (BOR), to provide those agencies with the hydrologic data needed for planning and operating water-resources projects. Additionally, some of the stations are funded by the USGS to support national programs of water-resources investigations; to collect data required by court decree, treaty, or compact; and to conduct hydrologic research. The USGS provides full support for fewer than 10 percent of the stations that it operates. Many of the stations funded primarily by State or local funds are critically important to USGS-funded programs, such as the National Water Quality Assessment (NAWQA) Program ( Leahy and Thompson, 1994). As discussed below, continuous streamflow data are essential to water-quality studies. The NAWQA Program could not be conducted without the stations funded by the Cooperative Program or other Federal agencies.

Figure 3.
Number of stations and sources of funds, 1994 fiscal year.

Because many of the stations are funded from multiple sources (Federal, State, and local agencies), each agency that participates in funding the stream-gaging program has a proprietary interest in the activity. State agencies, for example, view the data-collection activities in the Cooperative Program as a shared governmental responsibility in which they have a large, long-term financial investment and vested interest. The investment and the vested interest are carefully guarded, and changes in data-collection activities must be negotiated to mutual satisfaction. As a result of the strong vested interest, changes in the way the program is carried out require sensitivity to user reactions, thereby inhibiting unilateral action by the USGS.

Because interests in and the need for hydrologic data varies in time and space, stream-gaging networks are continually changing with time. The USGS attempts to balance availability of funding support with the needs of all interested parties to ensure that essential information is provided to all users. Budget constraints at State and Federal levels have forced many cooperators to reduce funding support for hydrologic data-collection activities. In some instances, monitoring activities at a particular site are discontinued because the needs of the supporting agency have been met. When funding support for a monitoring site is withdrawn, the USGS attempts to notify all potentially interested agencies of the impending changes to allow users of the data an opportunity to make alternative arrangements for funding the collection of data that are critical to their needs.

Uses of Streamflow Data

The USGS stream-gaging program provides hydrologic information needed to help define, use, and manage the Nation's water resources. The program provides a continuous, well-documented, well-archived, unbiased, and broad-based source of reliable and consistent water data. Because of the nationally consistent, prescribed standards by which the data are collected and processed, the data from individual stations are commonly used for purposes beyond the original purpose for an individual station. Those possible uses include the following:

Data for one or more of these purposes are needed at some point in time on virtually every stream in the country, and a data-collection system must be in place to provide the required information. The general objective of the stream-gaging program is to provide information on or to develop estimates of flow characteristics at any point on any stream. Streamflow data are needed for immediate decisionmaking and future planning and project design. Data, such as that needed to issue and update flood forecasts, are referred to as "data for current needs." Other data, such as that needed for the design of a future, but currently unplanned, bridge or reservoir or development of basinwide pollution control plans, are referred to as "data for future or long-term needs." Some data, of course, fit into both classifications; for example, a station that supplies data for flood forecasting and also provides data to define long-term trends.

Data for Current Needs

Streamflow data are needed at many sites on a daily basis for forecasting flow extremes, making water-management decisions, assessing current water availability, managing water quality, and meeting legal requirements. These activities require streamflow information at a given location for a specified time. These needs generally are best satisfied by operating a station to produce a continuous record of flow. The locations of the stations and the periods of operation are dictated by the uses to be made of the data.

More than one-half of the USGS stations provide current information (mostly by way of satellite telemetry) to agencies that operate water-resource systems and forecast floods. The NWS is charged by law with the responsibility of issuing forecasts and warnings of floods to the Nation to help save lives and to help mitigate property damage. The NWS uses data from USGS stations to forecast river stages and flow conditions on large rivers and their associated tributaries. Flood forecasts are issued at about 4,000 locations strategically located throughout the Nation. The reliability of flood forecasts depends on having reliable current data for precipitation and streamflow. The USGS collects the streamflow data, and the NWS collects the precipitation data and combines both types of data when making the flood forecasts. The NWS does not fund stations, but relies on the data from stations operated by the USGS for other agencies.

During the 1993 Mississippi River floods, USGS field personnel made more than 2,000 visits to stations in the flood-affected areas to verify that the instruments were working properly, to make repairs as needed, and to make direct measurements of the streamflow. Data from these stations were provided continuously to the NWS and the COE and formed the basis for flood forecasts that allowed people to be evacuated from areas about to be inundated. The COE and local agencies used the streamflow information to protect lives and property and to focus flood-fighting activities where they were most needed. As a national organization, the USGS was able to move staff from other offices into the disaster areas. Because these hydrologists and technicians were already familiar with the equipment and procedures, they could begin to work immediately upon arrival in the area. This same experience with the realtime use of USGS streamflow data is repeated several times each year as catastrophic floods strike various sections of the Nation.

Data for Future or Long-Term Needs

The collection of data to meet future needs often represents a larger challenge than does collection of data for current needs because the future needs are seldom known precisely and, in fact, may be impossible to anticipate. Because operating stations at all points on all streams is physically and economically impossible, mechanisms must be available to transfer streamflow information from stations to points where there are no streamflow data (ungaged sites).

Transfer of streamflow information for unregulated streams may be accomplished in many ways, ranging from the simple to the complex. Simple methods are interpolation between or extrapolation from gaging points on the same stream on the basis of drainage-area size. More complex methods may involve transferring information from basins with similar hydrologic characteristics, mapping station data to define approximate lines of equal runoff values, or correlating short records with long records. A statistical technique known as multiple-regression analysis has proven to be effective for defining equations (mathematical models) that relate streamflow characteristics to the basin and climatic characteristics that affect streamflow. The resulting equations usually are referred to as "regional relations" because they can be applied to ungaged streams within a defined hydrologic area or region. An example of a regional relation for estimating flood discharges for central Ohio is as follows ( Koltun and Roberts, 1989):

Q50 = 148 A^0.757 S^0.276 (St+1)^-0.355,


The above equation was computed by using values of Q 50, A, S, and St at 180 stations in central Ohio. The streamflow characteristic, Q50, was computed at each station by using streamflow records, and the basin characteristics A, S, and St were measured from topographic maps. To estimate Q50 at an ungaged site, the user determines the values of A, S and St for a specific site of interest from a topographic map and substitutes the values in the above equation. A compilation of regional relations for estimating flood discharges (like Q50) for rural and urban streams throughout the United States was given by Jennings and others ( 1994).

Studies of the uncertainties of these regional relations have been used to guide the USGS and its cooperators in determining how to change the stream-gaging program to reduce the uncertainty in estimates of streamflow characteristics. These studies permit the analyst to evaluate ways of reducing the uncertainty in the regional relations by adding new stations with certain ranges of basin characteristics, continuing operation of existing stations, or some combination of both approaches ( Medina, 1987).

Regardless of the methods used to transfer information, actual streamflow data are required. The stations that supply these data must be representative of the streams in the region. The data provided serve as the basis for defining and calibrating the equations (models) that serve as the transfer mechanism.

Some applications of data require long-term records to achieve a specified accuracy. The natural variation that is inherent in the flow of rivers produces uncertainty in estimates of the characteristics of those flows. The uncertainty is dependent on the variability of streamflow in the region and the length of streamflow record; uncertainty decreases as the record length increases. This is true no matter what is being discussed; for example, flood characteristics or the long-term average flow of the river. The relation between the standard error of estimate (a measure of uncertainty) and the record length for the mean-annual flow and the 50-year flood for Minnesota is shown in figure 4. If errors are normally distributed, then the standard error of estimate is the error to be expected for about two-thirds of the streamflow estimates.

Figure 4.
Relation between standard error of estimate and record length for Minnesota (from Benson and Carter, 1973).

The relation in figure 4 shows that given a 20-year record at a station, the 50-year flood can be estimated for that site with a standard error of about 35 percent. As the record length increases, the standard error or uncertainty in the 50-year-flood estimate decreases. When streamflow characteristics from stations are used to define a regional relation for use at ungaged sites, the error in the streamflow characteristics is a part of the total error in the regional relation. For example, note the scatter around the regional (regression) relation between the 50-year flood and the drainage area for rural streams in eastern Massachusetts ( Wandle, 1983) (fig. 5). The scatter about the regional relation includes the error that results when drainage area alone is used to estimate the 50-year flood, as well as the error in estimates of the 50-year flood at the individual stations. Thus, it is imperative that the data used in defining the equations be as accurate as possible, and that can be achieved only with long records. Note in figure 5 that a tenfold increase in drainage area results in about a fivefold increase in flood size. This is typical for flood characteristics, although the specific relation varies with hydrologic region.

Figure 5
. Relation between drainage area and 50-year flood for small rural streams in eastern Massachusetts (from Wandle, 1983).

Trend analysis is another application that requires long records. Concern is widespread that increased greenhouse-gas concentrations in the atmosphere are affecting the climate and the hydrology of the Earth. Analysts have used actual streamflow records to determine whether streamflows are beginning to change as a result of human activities or global warming. Natural climatic episodes of wetter or dryer than normal and lasting longer than a decade have been observed. Given the occurrence of such episodes and the inherent variability of streamflow, record lengths of more than 50 years are essential if real trends are to be detected. Slack and Landwehr ( 1992) reviewed the USGS data base to identify streamflow records that reflected natural conditions and could be useful in trend analysis. They identified 1,659 stations that could be used for this purpose in the United States and its possessions. The distribution of record lengths for these stations is shown in figure 6. More than 500 stations identified by Slack and Landwehr ( 1992) have record lengths in excess of 50 years.

Figure 6.
Number of stations and record lengths with acceptable data for studying climate fluctuations (from Slack and Landwehr,1992).

Specific Categories of Use

A recent nationwide evaluation of the USGS stream-gaging program identified uses of the data for individual stations in the program ( Thomas and Wahl, 1993). Between 1983 and 1988, uses of data were defined for 6,238 of the approximately 7,000 stations then operated by the USGS. Individual stations were identified as belonging to one or more of nine categories on the basis of the principal uses made of the data. The uses of data were determined through a survey of cooperators and other known users of the data. These users were recognized as representing only a limited sampling of all users of streamflow data. Many other organizations and individuals use data from the stream-gaging program, but these uses cannot be easily documented. Many times those users (and uses) become known only when a station is discontinued.

Hydrologic systems.---One of the more common uses of streamflow data is to account for and monitor the flow through a river basin or to define the general hydrologic conditions in the basin. Development of water resources has so altered the hydrology of some streams that station data at a given point primarily reflects the human manipulations. Data from about 4,200 stations operated by the USGS are used to understand and evaluate the resource, diversions, and return flows (water that has been used for some application and is being returned to the stream) that must be accounted for. Data from these stations also are useful in estimating hydraulic characteristics of aquifers, ground-water recharge, and evapotranspiration and in calibrating ground-water models. At State and interstate levels, many of the stations serve a key role in the process of allocating and regulating water rights. These stations provide data to satisfy current and future needs. Regional hydrology.---Stations supplying data that are largely unaffected by manmade storage or diversion furnish much of the data needed for future or long-term needs. Because they provide data that reflect natural conditions, these stations serve as the basis for defining the characteristics of streamflow and for developing the regional relations described in a previous section of this report. Data from about 3,800 stations operated by the USGS can be used for this purpose. Designers and planners of water-control and water-related facilities increasingly use the statistical characteristics of streamflow rather than the flow for specific periods in the past. For example, many highway bridges are designed on the basis of the flood that will be exceeded on the average of once in 50 or 100 years. Determining the appropriate design flood for a highway bridge is critical to balancing construction costs against risks to human safety and potential damage to property. Using too small a design flood can lead to a bridge design that causes water to back up and inundate the road itself or property along the flood plain upstream from the bridge. Too large a design flood can lead to a design that is wasteful and requires an unnecessarily wide opening, an unnecessarily high roadway, or both. By using long-term streamflow records, storage reservoirs can be designed on the basis of the probability of deficiency of storage to meet given discharge rates from storage. The water available for dilution of treated wastewater releases or other similar purposes may be stated in terms of the mean flow or probability of nonexceedance of flow magnitudes for periods of a year, season, month, week, or day. For example, if estimated low flows are understated, then there would be a requirement for additional costly wastewater treatment to meet water-quality standards. However, if low flows are overstated, then the treatment requirements would lead to unacceptable water quality when low flows occur.

Project operation.---Data from stations in this classification are used on an ongoing basis to assist water managers in making daily operational decisions. These decisions include managing daily and hourly flows through gates and hydropower penstocks, pumping water into diversion systems or hydroelectric reservoirs. They also include extensive balancing of uses among multiple sources of water in regional systems, including many reservoirs and ground water supplies. Such decisions are a daily reality when dealing with water requirements for municipal, industrial, and agricultural uses; hydroelectric power generation; and space for flood control in reservoirs. For example, data from about 2,900 stations operated by the USGS are used by the COE, the BOR, and others to operate more than 2,000 flood control, navigation, and water-supply reservoirs. Data from stations in this category satisfy a current need.

Hydrologic forecasting.---Data from stations so classified provide information for flood and water-supply forecasting. These stations play a key role in efforts by Federal, State, and local agencies to protect the lives and welfare of the general public through the evacuation of people from areas about to be inundated. More than 3,000 of the stations operated by the USGS are used in the NWS's flood-forecasting system. These data supply an immediate and high-priority current need. Because these stations are located at critical points on streams, they also generally provide valuable information on the statistics of flows that will be quite useful for meeting future needs.

Water-quality monitoring.--- The stations discussed in this paper are only a part, albeit the largest part, of the USGS hydrologic data-collection program. Other program components provide data on the chemical quality of water resources, sediment in streams and lakes, surface- and ground-water resources, and water use (fig. 7). Although the various program components are funded separately, they are highly interdependent and complementary. The programs on water quality and sediment provide information on the concentrations of chemical constituents in the water. The sediment and chemical quality of a river is intimately linked to the streamflow. Rapid variations in streamflow due to rainfall or snowmelt typically are associated with rapid variations in sediment or chemical concentrations. Consequently, understanding the movement of sediment and chemicals in a river depends on the availability of water samples at these times of rapid flow variation. One of the ways this is accomplished is to equip the station with an automatic pump sampler that is activated by a microcomputer programmed to call for samples based on stage, changes in stage, concentrations of chemical constituents, time since the last sample, or some combination of the above. These automatic sampling systems are vital to the study water-quality impacts of urban or agricultural land uses in small watersheds. The stations in this category also provide the flow data required to convert concentrations to loads (the total amount of the material transported by the water). The load transported by the flow is needed to understand fully and monitor the movement and fate of the material in flowing water. The approximately 2,700 stations operated by the USGS that provide discharge data for water-quality monitoring are fulfilling a current need. However, these data also may fulfill a future need if they are used to examine long-term trends in water quality or to determine the relative importance of various sources of pollution to a water body such as a reservoir, lake or estuary.

Figure 7.
Percentage distribution of funds for U.S. Geological Survey hydrologic data collection, 1994 fiscal year.

Planning and design.---Data from about 1,100 stations operated by the USGS in this category are needed to plan and design a specific project, such as a reservoir, levee, water-treatment facility, or hydroelectric powerplant. Because these data relate to a specific project, they generally are filling a current need.
Legal obligations.---Data from these stations satisfy a legal responsibility of the USGS or of signatories of treaties, compacts, and decrees. The USGS operates about 250 stations in support of 17 interstate compacts, 2 Supreme Court decrees, and 1 international treaty. Research.---Data from about 700 stations operated by USGS are collected for a particular research or water-investigation study. As such, the data supply a current need. The length of time that the data will be needed is dictated by the particular project. Some research needs, such as detection of hydrologic trends, can be met only by long-term, high-quality streamflow records.
Other.---These stations supply data for uses that do not fit into any of the eight categories above. These include, for example, recreational purposes, such as providing data for canoeists, rafters, and fishermen. Data from about 700 stations operated by the USGS supply a current need for water-resource information.

A growing number of stations are used for purposes that do not fit readily into one of the above nine categories. Data needed to define instream uses are good examples. Instream use refers to water that is used, but not withdrawn, from a surface-water source. Instream uses can be broadly characterized as streamflows required to meet human, ecological, or environmental needs. Human needs include recreation, hydroelectric power generation, transportation, waste assimilation, aesthetics, and cultural-resource preservation. Ecological or environmental needs include fish and wildlife habitat, wetlands preservation, freshwater dilution of saline estuaries, and maintenance of the riparian zone. Thus, these uses cut across most of the nine categories discussed above.

Quantitative estimates for most instream uses are difficult to compile. However, because such uses compete with offstream uses and affect the quantity and quality of water resources for all uses, effective water-resources management requires that methods, definitions, and procedures be devised to enable instream uses to be assessed quantitatively. The need to maintain some flow in streams has long been recognized as an important requirement for healthy stream ecosystems. In recent years, many court and compact decisions also have recognized the importance of instream flows and often have mandated an increase in instream flows to meet various environmental, recreational, and water-quality needs. Data from stations are critical to determine whether mandated instream flows are being maintained.

History and Growth of the Stream-Gaging Program

The USGS, which was created in 1879, was authorized in 1894 to survey irrigable lands in arid areas and to measure the flow of rivers and streams. As noted in an earlier section, the first USGS station was established on the Rio Grande in 1889. In 1895, the first Cooperative Program in the Nation began in Kansas through an agreement with the newly established Kansas Board of Irrigation Survey and Experiment (now known as the Division of Water Resources of the Kansas Department of Agriculture). This agreement provided for measurement of streamflow at seven sites to ascertain water-supply potential. In 1995, 100 years after the inception of the Cooperative Program, the USGS operates 166 stations in Kansas, 84 of which are operated in cooperation with 10 State, city, or local agencies (fig. 8). The other stations are supported by either Federal agencies or funds appropriated to the USGS.

Figure 8.
Areal distribution of stations for Kansas, by funding source.

The initial growth of the stream-gaging program was slow. At the turn of the century, only 163 stations were in operation. Most of the stations were in the West and were used to satisfy needs for irrigation. Growth of the program after 1900 was more dramatic, as shown by the number of active stations in each decade from 1900 to 1990 (fig. 9).

Figure 9.
Number of stations in the U.S. Geological Survey data base, 1900--90.

The growth and evolution of the USGS stream-gaging program was related to increased concern about floods and droughts, the increased use of water for irrigation and hydroelectric power, and specific legislative acts. The Federal Power Act was passed in 1920; during the next 20 to 30 years, planning for hydroelectric power development caused increased need for data. Congress passed legislation in 1929 that officially recognized the Cooperative Program in which costs are shared with State and local agencies, and in the ensuing years, cooperative stream-gaging programs were established with many State and local agencies. Also, the severe midcontinent drought in the early 1930's and the floods in 1936--37 in the Ohio and the Potomac River Basins increased the awareness among Federal, State, and local agencies that management of the water resources requires comprehensive, reliable streamflow data.

Passage of the Watershed Protection and Flood Prevention Act of 1954 and construction of the interstate highway system in the 1960's increased the need for streamflow data for small watersheds. Some of this need was provided by partial-record stations that recorded data only for flood peaks, but the numbers of continuous-record stations also increased. The need for data at the thousands of points where the highway systems crossed streams created an immediate need for methods to estimate flood magnitudes at ungaged sites. This need was satisfied by streamflow data to calibrate the regional equations used to make those estimates. The National Flood Insurance Act of 1968 increased emphasis on flood-plain mapping and emphasized the need for reliable flood-frequency data.

The Surface Mining and Control and Reclamation Act of 1977 also increased the need for data on streams affected by surface mining and other energy development. However, the additional stations constructed in the 1970's generally were offset by reductions in other areas of the network. Reevaluation of the program in the early 1970's ( Benson and Carter, 1973), the beginning of stringent financial constraints on the parties to the program, and the completion of many water-development projects were factors in limiting expansion of the program.

The major factors that have affected trends in the network since the late 1970's appear to be related to economic concerns and energy programs. In the 1970's, the oil crises gave impetus to a large expansion of research in coal and oil shale as sources of energy. Definition of the effects of such energy development on streamflow and water quality required streamflow data. Numerous stations were installed to provide those data. As concern for energy sources waned in the early 1980's, many of those stations were discontinued.

In 1987, a poll was made of USGS offices to identify stations discontinued or started from 1981 to 1986. This poll was taken in response to NWS concerns that the number of stations in the USGS stream-gaging program was declining. Between 1981 and 1986, 873 stations were added, but 1,744 stations were discontinued; thus, there was a net loss of 871 stations from the program. This illustrates the complexity of change; stations that were added and then deleted during the period (125) were not counted in this poll.

A more recent poll of USGS offices in Delaware, Georgia, Iowa, Idaho, Maryland, Michigan, Oklahoma, and South Dakota showed little net change in numbers of stations between 1985 and 1994. These States, which were selected as a representative sample, had 832 stations at the beginning of the period. During the period, 189 stations were added, and 170 were dropped; this represents about a 20-percent turnover rate in 10 years. An additional 29 stations were started and dropped. Of the stations that were dropped, record lengths were more than 20 years at 97 stations and more than 40 years at 30 stations.

Continue to ' Data-Collection Process ', or return to ' Table of contents '
Stream-Gaging Program of the U.S. Geological Survey
Reston, Virginia, 1995
By Kenneth L. Wahl, Wilbert O. Thomas, Jr., and Robert M. Hirsch

URL for this page is <URL=>