Tracing Pollution Sources in an Urban Watershed: A GIS-Based Predictive Model of Bacteria Contamination in Stormwater

DSpace/Manakin Repository

Tracing Pollution Sources in an Urban Watershed: A GIS-Based Predictive Model of Bacteria Contamination in Stormwater

Citable link to this page

 

 
Title: Tracing Pollution Sources in an Urban Watershed: A GIS-Based Predictive Model of Bacteria Contamination in Stormwater
Author: Hrycyna, Andrew
Citation: Hrycyna, Andrew. 2016. Tracing Pollution Sources in an Urban Watershed: A GIS-Based Predictive Model of Bacteria Contamination in Stormwater. Master's thesis, Harvard Extension School.
Full Text & Related Files:
Abstract: This project explores whether it is possible to predict which urban stormwater pipe networks are most likely to be contaminated by wastewater inputs based on geographic information about the areas they drain. Wastewater pollution introduced into urban stormwater systems is a major source of impairment of water bodies in the United States, introducing pathogens and other pollutants, into rivers, streams, and lakes. Recent stormwater permits require extensive testing for bacteria. Efficiencies could be gained in remediating problems if an evidence-based prioritization scheme could target stormwater pipe networks based on publicly available information. I use a large data set of bacteria data in stormwater from the Mystic River watershed in Massachusetts, along with a GIS methodology, to explore a hypothesis that some features of the stormwater networks and the land they drain can usefully predict which networks will exhibit high bacteria values. Multiple regression analysis shows that pipe length, population density, and age of buildings in an area are significant predictors of high bacteria concentrations in the Mystic River dataset. In addition, I use the final regression model to estimate bacteria loads from stormwater outfalls. I conclude that the evidence supports a pollution-tracking prioritization scheme that tests large pipe networks first, at a minimum. I discuss the possible reasons for this somewhat surprising result, and suggest further ways to extend and refine this modeling approach.
Terms of Use: This article is made available under the terms and conditions applicable to Other Posted Material, as set forth at http://nrs.harvard.edu/urn-3:HUL.InstRepos:dash.current.terms-of-use#LAA
Citable link to this page: http://nrs.harvard.edu/urn-3:HUL.InstRepos:33797346
Downloads of this work:

Show full Dublin Core record

This item appears in the following Collection(s)

 
 

Search DASH


Advanced Search
 
 

Submitters