USING MACHINE LEARNING TO INVESTIGATE THE ROLE OF CAFO EXPANSION ON NUTRIENT POLLUTION TRENDS IN EASTERN NORTH CAROLINA

Agriculture and wastewater practices are major contributors to nutrient loading and coastal eutrophication issues globally. The lower Cape Fear River basin in eastern North Carolina, USA, is particularly affected, with one of the highest densities of livestock production facilities, known as concentrated animal feeding operations (CAFOs), in the world. Long-term water quality monitoring by the Lower Cape Fear River Program revealed chronic nutrient and fecal bacteria pollution and increased trends over the past 20 years with hotspots in CAFO-dense watersheds. While swine CAFO density has remained stable since the 1990s, poultry CAFOs have rapidly expanded, yet long-term geospatial data for poultry CAFOs was unavailable. The growth of the poultry CAFO industry is a high priority research need to understand the impacts on water quality and human health in this already CAFO-dense area. A machine-learning image classification model was developed to identify and quantify poultry CAFOs constructed in North Carolina over the past 20 years using public aerial imagery from the National Agricultural Imagery Program and processed with Google Earth Engine and Google Colaboratory. Combined with long-term water quality data, the resulting dataset fills a critical data gap and provides a framework for assessing the long-term impact of CAFOs on water quality. The findings demonstrate the utility of machine learning for monitoring pollution sources and provides insights for environmental management and policy decisions in the region.

Presentation Preference: Oral

Primary Presenter: Colleen Brown, University of North Carolina Wilmington (cnb6822@uncw.edu)

Authors:

Colleen Brown, University of North Carolina Wilmington  (cnb6822@uncw.edu)

Michael Mallin, University of North Carolina Wilmington  (mallinm@uncw.edu)

Yang Song, University of North Carolina Wilmington  (songy@uncw.edu)