Date of Award

Spring 1-1-2018

Document Type


Degree Name

Doctor of Philosophy (PhD)

First Advisor

Michael P. Hannigan

Second Advisor

Marina E. Vance

Third Advisor

Jana B. Milford

Fourth Advisor

Shelly L. Miller


Natural gas, as an energy resource can exert both positive and negative influences on air quality where people live, work, and go to school. Air quality in basins where oil and gas are produced from geologic formations can be potentially degraded by industry activities, but using natural gas in place of solid fuels like wood and coal for home heating and in other applications can potentially result in improved air quality. Low- cost gas sensors have emerged recently with great potential to help inform air quality on the scales that people live in ways that traditional instrumentation is not well suited, though the usefulness of these tools is complicated by cross sensitivity to environmental variables like temperature and humidity, as well as potentially confounding gas species. The ability of low-cost gas sensors to yield meaningful information about air quality, with relevance to human and environmental health, is therefore contingent on progress in terms of sensor signal quantification methods, best practices for experimental design and deployment, along with data quality assessment and interpretation.

In this dissertation, such methods are developed and applied, employing low-cost gas sensors to characterize air quality in both indoor and ambient environments, in the context of natural gas production and end use as a home heating fuel. Carbon monoxide (CO) measurements are used to characterize how home heating fuels can differentially influence air quality in homes on the Navajo Nation. CO levels in homes are quantified with uncertainty estimation and are employed to estimate air exchange rates in homes and CO emission rates. Methods to measure air quality in oil and gas production basins using arrays of low-cost gas sensors are also developed and analyzed. Field normalization sensor signal quantification methods employing both artificial neural networks and multiple linear regressions are compared. The sensitivity and robustness of each quantification method is explored for each gas species. To further understand how distributed grids of sensor measurements can inform spatial and temporal patterns of air quality in oil and gas production basins, the performance of these sensor quantification methods are assessed when extended to new sampling locations.