A land use regression model to predict emissions from oil and gas production using machine learning

Abstract

Emissions from oil and natural gas (O&G) wells such as nitrogen dioxide (NO2), volatile organic compounds (VOCs), and ozone (O3) can severely impact the health of communities located near wells. With the O&G industry growing and 17.6 million people living within a mile from an O&G well, the effect of O&G activity on residents is especially pertinent. In this study, we used O&G activity and wind-carried emissions to quantify the extent to which O&G wells affect the air quality of nearby communities, revealing that NO2, NOx, and NO are correlated to O&G activity. We then developed a novel land use regression (LUR) model using machine learning based on O&G prevalence to predict emissions. Many LUR models fail to account for O&G sources, therefore we hypothesized that the inclusion of O&G sources in land use regression models provides an increase in accuracy when predicting emissions. The model performed effectively for NO2, outperforming past LUR models which did not involve O&G activities. The model makes it possible for not only communities, but also families and individuals, to determine the effect that O&G has on their homes. With current modeling techniques failing to observe the effects of O&G in the face of the growing O&G industry in the U.S., it is crucial that the public is educated on the effect of the O&G industry on their daily lives and has the tools to monitor these effects.