Statistical Assessment of Violations of Water Quality Standards under Section 303(d) of the Clean Water Act

• Published in: Environmental science & technology Vol. 35; no. 3; pp. 606 - 612
• Main Authors: Smith, Eric P, Ye, Keying, Hughes, Chris, Shabman, Leonard
• Format: Journal Article
• Language: English
• Published: Washington, DC American Chemical Society 01.02.2001
• Subjects: Applied sciences; Bayes Theorem; Clean Water Act-US; Exact sciences and technology; False Negative Reactions; Federal legislation; Humans; Pollution; Quality Control; Reference Values; States; Statistical analysis; United States; United States Environmental Protection Agency; Water Pollutants - analysis; Water pollution; Water quality; Water resources; Water Supply; Water treatment and pollution; United States; North America; America; USA; Statistical analysis; Water quality; Error rate; Decision aid; Binomial distribution; Bayes methods; Environmental management; Environmental protection
• ISSN: 0013-936X; 1520-5851
• DOI: 10.1021/es001159e

Section 303(d) of the Clean Water Act requires states to assess the condition of their waters and to implement plans to improve the quality of waters identified as impaired. U.S. Environmental Protection Agency guidelines require a stream segment to be listed as impaired when greater than 10% of the measurements of water quality conditions exceed numeric criteria. This can be termed a “raw score” assessment approach. Water quality measurements are samples taken from a population of water quality conditions. Concentrations of pollutants vary naturally, measurement errors may be made, and occasional violations of a standard may be tolerable. Therefore, it is reasonable to view the assessment process as a statistical decision problem. Assessment of water quality conditions must be cognizant of the possibility of type I (a false declaration of standards violation) and type II (a false declaration of no violation) errors. The raw score approach is shown to have a high type I error rate. Alternatives to the raw score approach are the Binomial test and the Bayesian Binomial approach. These methods use the same information to make decisions but allow for control of the error rates. The two statistical methods differ based on consideration of prior information about violation. Falsely concluding that a water segment is impaired results in unnecessary planning and pollution control implementation costs. On the other hand, falsely concluding that a segment is not impaired may pose a risk to human health or to the services of the aquatic environment. An approach that recognizes type I and type II error in the water quality assessment process is suggested.