Page images

Cost-Effectiveness of the National Stream-Gaging Program

The U.S. Geological Survey began a nationwide analysis of its stream-gaging program during fiscal year 1983. The purpose of the analysis is to define and document the most cost-effective method of furnishing streamflow information. The analysis is being carried out over a 5-year period with about 20 percent of the program being analyzed each year. The Survey operates about 8,000 continuous-recording gaging stations nationwide that provide streamflow information for a large variety of users. These gaging stations will be evaluated (1) to identify the principal uses of the data and to relate these uses to funding sources, (2) to identify alternate, less costly methods of furnishing needed information, and (3) to define strategies for operating the program to minimize the standard error in streamflow data while staying within the operating budget. The first two steps are designed to ensure that sufficient need exists for operating a gaging station. The third step provides for allocation of financial and manpower resources among the stations that remain in the program after the screening process, so that the program is operated in the most cost-effective manner. An analysis completed for the State of Maine early in fiscal year 1983 will serve as a prototype.

In the first step, the known uses of streamflow data generated at a gaging station are compared against the objectives of the stream-gaging program to ensure sufficient justification exists for Survey involvement at that station. Deficiencies in the existing data-collection program are evaluated to ensure that all information needs are met. The responsiveness of the operation of each station to the types of uses also is evaluated to see that streamflow information is timely. For example, analysis of the data uses for 51 stations in Maine indicated that three stations should be discontinued as soon as is practical and that an additional three stations should be discontinued at the end of short-term projects. Analysis also indicated that, as funds become available, additional stations should be established in the interior of Maine to better define regional hydrology.

The second step of the analysis is used to determine whether sufficient streamflow information can be generated at a station

by methods other than operating a continuous-record station. Primarily, two alternate methods are considered, a flowrouting model and a statistical regression model. The flow-routing model uses the traveltime of flow between stations, the storage in the stream channel, and hydrologic routing techniques to transfer daily flows from an upstream station to a downstream one. The statistical regression model correlates daily flows at the station of interest with daily flows at other nearby stations. Once calibrated, both models can be used to estimate daily flows at discontinued stations by using daily flows from operating stations. The accuracy of the estimated streamflow must be suitable for the intended usage for an alternate method to be viable. In the Maine analysis, there was one station where both models provided daily discharges of sufficient accuracy for the intended usage. Both models were calibrated using all existing data, and the recommendation was to continue operating the station until sufficient data were available to verify the models. The final step is used to determine the best allocation of money and manpower among the stations that remain in the program after the two screening steps. Because there are so many uses made of streamflow data, minimization of the standard error of streamflow data, expressed as percentages, is chosen as the general measure of the program's effectiveness. This part of the analysis defines the uncertainty function for each station in the program, develops the necesary cost information, and determines the number of visits necessary to each station to minimize the uncertainty.

The uncertainty function relates the standard error of streamflow data to the number of visits and measurements made per year or season. Examples of typical uncertainty functions from the Maine study are given in figure 1. These uncertainty functions are computed using a statistical technique that evaluates the accuracy of the streamflow rating curve, the accuracy of transferring flows from nearby stations, and the variability of historical flows at the station. The rating curve at each station is the relationship that enables the hydrologist to convert the



Figure 1. Typical uncertainty functions for three gaging stations in Maine.

0 5


recorded water-surface elevation (gage
height or stage) to streamflow. At some
sites, additional correlative data are
necessary to determine the flow, such as
the fall in water surface between sites.
When the recorder at the station fails to
record the water-surface elevation (or
other correlative data), the rating curve
cannot be used, and daily flows must be
estimated from flows at nearby sites or
from historical flows at the station. The
uncertainty function includes the variability
or standard error of flows estimated in
these various ways.

Once the uncertainty functions for each
station are known, various costs associ-
ated with stream gaging can be deter-
mined. Feasible routes are defined for serv-
icing the stations, and each station is
assigned to one or more routes. The cost
of servicing each station, route costs, and
the minimum number of times each station
should be visited are determined. The fixed
costs of operating each station, including
the cost of computing records and their
storage and publication, are also deter-
mined. This information and the uncertain-
ty functions are input to a computer pro-
gram that determines the number of times
each route is used. The routes selected are
those with the largest reduction in uncer-
tainty per dollar of expenditure. By varying
the total budget and repeatedly running
the program, an uncertainty or average

10 15


standard error relationship with the budget can be developed. Figure 2 is an example of an uncertainty-cost relationship for Maine. The original Maine stream-gaging program, consisting of 51 stations, operated with an annual budget of $211,000. As a result of the data-use analysis, it was recommended that 6 of the original 51 stations in the Maine stream-gaging program be discontinued. The stream-gaging program analyzed for cost-effectiveness consisted of 45 stations. The current criteria for operating the 45-station program require a budget of $180,300. This is the circle in figure 2 marked "Current Practice." The average standard error of the streamflow records was 17.7 percent. As can be seen in figure 2, this overall level of accuracy could be maintained with a budget of about $ 170,000 if allocation of resources among the gages was altered. The recommendation was to modify the operation of the program and to use the residual $10,300 to increase receipt of data from the interior of the State. The relationship in figure 2 indicates the reduction in uncertainty that can be achieved by increasing the total budget.

Studies like the one in Maine are scheduled for completion in 17 States during fiscal year 1983. The entire stream-gaging program will be analyzed over the next 5 years as part of the continuing effort of the Geological Survey to evaluate the Na

[graphic][merged small][merged small][merged small][merged small][merged small]

Contamination of Ground Water by Coal-Tar Derivatives in St. Louis Park, Minnesota

Section showing major processes controlling the transport through ground water and fate of coal-tar derivatives, St. Louis Park area, Minnesota.

Operation of a coal-tar distillation and wood-preserving facility in St. Louis Park, Minnesota, from 1918 to 1972 resulted in severe ground-water contamination. In 1978, the U.S. Geological Survey began detailed studies of the transport and fate of coal-tar derivatives through ground water in the area. Local, State, and Federal agencies will use the results of the studies to guide management decisions and to design remedial action. The studies were conducted in cooperation with the Minnesota Department of Health, Minnesota Pollution Control Agency, city of St. Louis Park, and the U.S. Environmental Protection Agency.

The problem of most immediate concern to the city and to the State and Federal regulatory agencies is the presence of toxic organic compounds in water withdrawn from some municipal wells. When the first municipal well was drilled in 1932, the Prairie du Chien-Jordan aquifer contained water having a distinct coal-tar taste. The well is 3,500 feet from the plant site. From 1978 to 1981, use of seven more municipal wells in this aquifer was discontinued because the wells yielded water containing trace amounts of coal-tar compounds, including at times and places, the carcinogen benzo(a)pyrene.

The Prairie du Chien-Jordan aquifer is the region's major ground-water resource. About 75 percent of ground-water withdrawals in the St. Louis Park and Minneapolis-St. Paul metropolitan areas are from this aquifer. The aquifer has good natural protection from near-surface sources of contamination. In the St. Louis Park area, it is 250 to 500 feet below land

surface and is overlain by glacial drift, two bedrock confining beds (Glenwood and basal St. Peter), and two bedrock aquifers (Platteville and St. Peter). Nonetheless, it is now contaminated because materials entered the aquifer through at least five wells that hydraulically connect more than one aquifer (multiaquifer wells). The single major source is a well on the former plant site that was drilled in 191 7 to an original depth of 909 feet. When first geophysically logged by the Survey in 1978, the well had filled to a depth of 595 feet. The uppermost 100 feet of the fill was mostly coal tar. Moreover, approximately 1 50 gallons per minute of contaminated water was moving through the well bore from the St. Peter aquifer into the Prairie du Chien-Jordan aquifer.

Contaminants in the Prairie du ChienJordan aquifer have moved at least 2 miles northeast and southeast of the plant site. The direction and rate of contaminant movement changes with time because the bedrock ground-water flow system continually adjusts to hydraulic stresses caused by water withdrawals and flow through multiaquifer wells. Contaminants move rapidly through the Prairie du ChienJordan aquifer because the upper part is a carbonate rock with fractures and solution channels. Consequently, the concentration and composition of contaminants in water pumped from the Prairie du Chien-Jordan aquifer through individual industrial and municipal wells fluctuates with time.

Contaminants entered the uppermost bedrock aquifer, the Platteville, directly from the drift and moved at least 4,000 feet from the plant site. Locally, the con

[merged small][merged small][merged small][subsumed][merged small][ocr errors][merged small][merged small][merged small][graphic]

taminants have reached the St. Peter aquifer through the Glenwood confining bed and (or) through bedrock valleys where the confining bed has been removed by erosion.

The greatest mass of contaminants is in the drift near the plant site. Coal-tar derivatives reached the water table by percolation through the unsaturated zone and through ponds that received surface runoff and process water from the plant. Parts of the drift contain an undissolved liquid mixture of many individual coal-tar compounds. Chemical analyses of organic fluid and water from a monitoring well completed in the drift 50 feet below the water table identified more than 200 individual organic substances. The viscous organic fluid is denser than water and has moved slowly downward independent of the

direction of ground-water flow. Ground water entering the area of the plant site through the drift is contaminated by partial solution of the organic fluids and by release of compounds sorbed on the drift materials. The contaminated water moves laterally to the east and southeast and downward into the Platteville aquifer. Water in the drift 4,000 feet from the site contains less than 10 milligrams per liter of disolved organic carbon but has a distinct chemical odor and contains a large proportion of coal-tar compounds highly soluble in water.

One major group of coal-tar compounds (phenolic compounds) is being degraded to methane and carbon dioxide by bacteria that metabolize phenolic compounds in the anaerobic (oxygen-free) environment that exists in the aquifer. This finding is of

[ocr errors][graphic][merged small][subsumed][merged small]
« PreviousContinue »