Margin of Safety in TMDLs: Natural Language Processing-Aided Review of the State of Practice

Robert Nunoo, Paul Anderson, Saurav Kumar, Jun Jie Zhu

Research output: Contribution to journalArticlepeer-review

10 Scopus citations

Abstract

The margin of safety (MOS) accounts for uncertainties in the total maximum daily load (TMDL) development process and the variabilities involved in simulating systems, providing a complete description of the degree of protection for a waterbody. Despite numerous discussions on the estimation and incorporation of MOS in TMDLs, there are few post-TMDL studies about the MOS selection process, and there are limited guidelines for selecting MOS values. In this study, natural language processing was employed to review MOS values of TMDLs approved between 2002 and 2016. Reasons such as type of impairment, waterbody types, and designated waterbody given by TMDL developers for chosen MOS values uses were explored. MOS values across states and United States Environmental Protection Agency regions were also analyzed and compared. The results suggested the MOS value of 10% of the estimated load capacity of a waterbody is the most used value across the states and territories of the United States and that 84% of the explicit MOS values selected were not based on any uncertainty estimation method. In addition, the waterbody type and the designated water-use category appear to be correlated with MOS values, and lakes and designated use for aquatic life protection generally had larger MOS values.

Original languageEnglish (US)
Article number04020002
JournalJournal of Hydrologic Engineering
Volume25
Issue number4
DOIs
StatePublished - Apr 1 2020
Externally publishedYes

Keywords

  • Machine learning
  • Margin of safety
  • Natural language processing
  • Total maximum daily load (TMDL)
  • Uncertainty

ASJC Scopus subject areas

  • Environmental Chemistry
  • Civil and Structural Engineering
  • Water Science and Technology
  • General Environmental Science

Fingerprint

Dive into the research topics of 'Margin of Safety in TMDLs: Natural Language Processing-Aided Review of the State of Practice'. Together they form a unique fingerprint.

Cite this