Pollution in the Press: Employing Text Analytics to Understand Regional Water Quality Narratives

Abstract:

Water has and will continue to be at the foundation of our nation’s wellbeing. However, a significant portion of its infrastructure does not meet national standards. Nearly 40% of the United States population obtained their water from drinking water systems that did not meet federal regulations between 2016 and 2019. Regional variations in contaminants, water sources, and financial resources complicate causal analyses and understanding of community water systems’ health. Studies suggest a correlation between local awareness and community engagement in water quality issues, but how does one collect and analyze the relationship between community discussions and local water quality? By harnessing the power of natural language processing, we analyze a corpus of 1.8 million water-related articles from 37 local news sources across the United States to understand the narratives of water quality issues. Specifically, we assess the relationship between media coverage and water quality exceedances in communities as well as the influence of social, political, and economic factors on the coverage of water quality parameters. Preliminary results indicate regional variations in discussed contaminants, with lead discussed more frequently in the Northeast and Midwestern regions, while uranium is discussed more frequently in the West. These findings provide a basis for understanding variations in local discourses and associated implications for water quality-related security discussions.