• Background
• Purpose and Scope
• General Investigative Approach
• Selected Water Quality Constituents
• Source of Data
• Significant Findings

• STORET Database Access
• NWIS Web

Summary of Available State Ambient Stream-Water-Quality Data, 1990-98, and Limitations for National Assessment

In 1972, the Clean Water Act and subsequent amendments set goals for eliminating the discharge of contaminants into the Nation’s waters, provided Federal funding for construction of municipal sewage-treatment plants, directed the States to set water-quality criteria for waters within their boundaries, and funded development of programs to control nonpoint sources of contamination. The act also directed the States to submit biannual reports to the U.S. Environmental Protection Agency (USEPA) that presented a comprehensive water-quality assessment (305(b) report) and listed the water-quality impaired (degraded) streams or stream segments (303(d) report) within the States. In response to the provisions of the Clean Water Act, States established, modified, or enhanced their ambient stream-water-quality monitoring networks to provide the data necessary to assess stream water quality relative to State-established criteria. Data from these monitoring networks have been used to produce the biannual 305(b) and 303(d) reports to USEPA and, ultimately, to the Congress in the form of National Water Quality Inventory reports (compilation of State 305(b) reports).

State networks have been operated to meet specific goals and objectives established by the individual States. However, a national summarization of water quality can be accomplished in one of two ways. One approach would require that the States design and execute their individual monitoring approaches using identical designs so that the results can be summed to provide a national assessment. The other would be an analytical approach (some combination of statistical and deterministic models) that combine the 50 data sets in a manner that produces inferences about water quality that take into consideration the different sampling designs. Neither of these requirements has been met to date, and thus, the data are not suitable for national assessment or State-to-State intercomparisons, even though the data sets may be highly useful for purposes of State assessments and management.

The U.S. General Accounting Office (USGAO, 2000) examined USEPA’s National Water Quality Inventory reports and the adequacy of water-quality data for key water-quality decisions required by the Clean Water Act. The USGAO concluded that assessment data in State 305(b) reports are incomplete because the data do not represent all of the States’ waters. Additionally, USGAO identified substantial differences among States in virtually every aspect of monitoring and assessment. These differences stem from a combination of factors including variability in the (1) water-quality criteria (including designated or beneficial uses) for determining which water is degraded; (2) types of monitoring practices used to ascertain whether or not the criteria are exceeded; and (3) procedures used to assess water-quality data for determining if water-quality conditions are degraded. Much of this variability is the result of the flexibility allowed the States by the Clean Water Act to decide which streams to assess, what uses to designate for streams, and what water-quality criteria to assign for the designated uses.

Targeted, statistically based (probabilistic), and geographically based sampling designs are used in some State monitoring programs to provide the water-quality information required by the Clean Water Act to evaluate potential stream-water-quality degradation relative to designated uses and appropriate water-quality criteria. However, the variability in network designs and operational characteristics, such as the distribution of sites and frequency of sampling, among the States inhibits the use of available information for comparison of results among the States.

Because of the aforementioned differences in State ambient stream-water-quality monitoring networks and programs, water-quality officials in Kansas generally questioned the validity of previous national water-quality comparisons produced on the basis of information presented in State 305(b) and (or) 303(d) reports and, specifically, previous summaries of water-quality conditions in Kansas. In 2001, the U.S. Geological Survey (USGS) in cooperation with the Kansas Water Office (KWO), supported in part by the State Water Plan Fund, conducted an investigation of the comparability and representativeness of State ambient stream-water-quality databases for assessment of national water-quality conditions. Although it was beyond the scope of this investigation to evaluate the adequacy of the network designs, field and laboratory activities, and quality-assurance practices used in the various State monitoring programs, this investigation did provide additional information to support or contrast with previous investigations by the USGS that indicated data in many electronically accessible databases are of unknown quality and hydrologic context. For example, a previous comparison of data availablity in Colorado and Ohio indicated that fewer than 15 percent of water-quality analyses met the screening criteria judged necessary for the data to be included in a consistent database suitable for addressing broad-scope water-quality questions. Also, it was determined previously that for 12 evaluated water-quality constituents an average of only 22 percent of stream-water-quality monitoring sites in Colorado and 8 percent in Ohio had 10 or more analyses with concurrent streamflow data during 1977–84 and less than 10 percent of the stream-water-quality monitoring sites in Colorado and Ohio yielded sufficient data for determining changes in water-quality conditions. Concurrent streamflow data are necessary for hydrologic assessment such as water-quality trend analysis and constituent transport (load) calculations.

The purpose of this investigation was to produce a summary of available State ambient stream-water-quality data, 1990–98, and to examine some of the characteristics of the data that would limit its use in producing a national evaluation of stream water quality. Ambient stream-water-quality data are information collected by an integrated activity for the purpose of evaluating the physical, chemical, and biological character of water ecosystems in relation to human health, environmental and ecological conditions, and designated uses. Therefore, “ambient” refers to data collected for those purposes and does not mean or is it used to indicate pristine water-quality conditions. On the contrary, States may have established monitoring sites in their ambient networks for many reasons to include the documentation of stream-water-quality effects from human-related sources. These sources may be both point and nonpoint sources of contamination. Water-quality conditions determined in many or all State ambient stream networks probably are biased toward the water-quality effects generated from human-related activities because of the interest in these effects and their general pervasiveness.

The scope of this project was limited to States with electronically available water-quality data for sampling sites in their ambient stream-water-quality monitoring networks for 1990–98 (a maximum of 42 States). Water-quality data were statistically summarized by State.

The water-quality data used for this investigation were from monitoring sites representing each State’s ambient stream-water-quality monitoring network for a common time period. No attempt was made to normalize for how and why monitoring sites were selected such as targeted or probabilistic-based selection, the number of sites within each State, spatial distribution of sites, or temporal distribution of streamflow samples collected at those sites. A general schematic of the procedures used and assessment steps taken in the conduct of this investigation is presented.

Water-quality constituents selected for summary in this investigation were those most commonly responsible for water-quality degradation in the Nation’s streams. In preparation for the 1998 USEPA National Water-Quality Inventory Report to Congress, States and tribes assessed 840,402 mi of rivers and streams in the Nation and determined that 35 percent of those miles (291,263 mi) were degraded with respect to some water-quality factor. Of the degraded river and stream miles in 1998, 38 percent were degraded by sediment/siltation, 36 percent by fecal bacteria, and 29 percent by nutrients (species of nitrogen and phosphorus). Smaller percentages were degraded by trace metals and pesticides.

The data summary period for this investigation was 1990–98. The main source of water-quality data was the USEPA STORET Legacy Data Center. Of the 42 States with electronically available water-quality data, 28 State databases were acquired from the Legacy system. Data were acquired for five States (Colorado, Montana, New Hampshire, North Dakota, and Utah) from the modern STORET system that currently (2004) contains the 1990–98 data, and from State-located databases for three States (Maryland, Oklahoma, and Virginia). The USGS National Water Information System (NWIS) served as the primary or ancillary source of water-quality data for seven States (Connecticut, Idaho, Missouri, New Jersey, New Mexico, Rhode Island, and Wyoming). These seven States had cooperative agreements with the USGS to collect and analyze all or part of the water-quality samples for their ambient stream-water-quality monitoring networks.

Eight States (Alaska, Arizona, California, Hawaii, Maine, Massachusetts, North Carolina, and Vermont) did not have electronically available data. Alaska and Hawaii did not have ambient stream-water-quality networks during 1990–98. Arizona, Massachusetts, and North Carolina had ambient stream-water-quality networks, but chemical and physical data were not stored in the STORET system or in a State-located database that was electronically accessible. Maine and Vermont conducted only biological water-quality assessments; therefore, they did not have data for the chemical and physical characteristics summarized during this investigation. California did not have a centralized, electronically accessible State database. Ambient stream-water-quality responsibilities in California were divided among nine Regional Water-Quality Control Boards, each of which varied as to amount, availability, and accessibility of data.

• The electronic availability of data varied substantially among the States.

  • For example, forty-two States had pH data from ambient stream monitoring sites, whereas only 26 States had total arsenic data. Electronically accessible data were not available for Alaska, Arizona, California, Hawaii, Maine, Massachusetts, North Carolina, and Vermont.

• Comparisons of monitoring data among States are problematic for several reasons.

  • These reasons include the basic spatial design of monitoring networks; water-quality constituents for which samples are analyzed; water-quality criteria to which constituent concentrations are compared; quantity and comprehensiveness of water-quality data; sample collection, processing, and handling; analytical methods; temporal variability in sample collection; and quality-assurance practices. Large differences among the States in number of monitoring sites precluded a general assumption that statewide water-quality conditions were represented by data from these sites. Furthermore, data from individual monitoring sites may not represent water-quality conditions at the sites because sampling conditions and protocols are unknown. Because of these factors, a high level of uncertainty exists in the results of this investigation.

• National monitoring is, in reality, a compilation of 50 State networks with varying degrees of comparability.

  • Differences among the States in water-quality criteria and methodologies used in data assessment and decisions on water quality restrict comparability among the States and limit the consistency needed to produce an accurate national assessment of ambient stream-water quality.

• An analysis of the extent and comprehensiveness of State ambient stream-water-quality networks indicated that these networks may be lacking in the diversity of collected data and that many monitoring sites in those networks have few samples.

  • Many basic water-quality constituents do not have the number of analyses recommended (30) by the U.S. Environmental Protection Agency to make appropriate water-quality and total maximum daily load decisions.

• Variability in network designs and operational characteristics among the States prevented the use of available information for a scientifically defensible comparison of results among the States.

  • Targeted, statistically based (probabilistic), and geographically based sampling designs are used in some State monitoring programs to evaluate potential stream-water-quality degradation relative to designated uses and appropriate water-quality criteria. When appropriate designs are used, this information also may be used to describe cause-and-effect relations, seasonal variability and time trends, mass transport, and as an aid to management decisions.