"The primary source of compliance information in water quality monitoring is self-reported data. Despite the heavy reliance on self-reported data in United States environmental regulation, the U.S. General Accounting Office has expressed concerns regarding the potential for fraud in environmental self-reports. Furthermore, recent research indicates that the methods used by state enforcement are unlikely to detect fraud. Therefore, the need for data-driven methods to support regulatory enforcement is an important area of research. In this thesis, we evaluated the applicability of data-driven methods for assessing compliance of wastewater treatment plants (WWTP) self-reported datasets based on a description of the variability in these data streams. For this purpose, first a literature review was conducted (1) to determine the goals of the Clean Water Act programs; (2) identify limitations of current monitoring efforts and data gaps in the understanding of the sources of variability in WWTPs data; and (3) to identify appropriate predictive analytical methods to address the problems. Second, the applicability of Benford's Law as a method for uncovering irregularities in the distribution of first and second digits in a sample dataset was tested and its effectiveness was discussed. Finally, the use of other promising approaches, which may be capable of finding mishandling in wastewater treatment plants are presented with preliminary data."--Page ii.
Read
