TowerScout Adaptation – Automated Image Analysis to Identify Cooling Towers

About the Program

Category: Information Technology (Best in Category)

Year: 2023

The CDC, US National Academy of Science, and the World Health Organization, among other health and scientific organizations, recognize cooling towers pose a significant potential for dispersion of Legionella-containing water mist and are associated with some of the largest outbreaks of Legionnaires’ Disease. Accurate knowledge of cooling tower locations is critical to prevent and respond to public health emergencies, however the manual and time-intensive process of identification has long been a barrier to development of a comprehensive inventory. The Los Angeles County Enterprise GIS (eGIS) team and Department of Public Health collaborated to adapt TowerScout, a software platform that uses machine learning to automatically detect cooling towers visible from aerial imagery, to analyze the County’s high-resolution (10cm) commercial aerial imagery and build an initial dataset of likely cooling tower locations across the County. Utilizing Python programming, geospatial tools and methods, and the TowerScout machine learning model, the team developed an efficient framework and workflow for automatically detecting potential cooling tower locations and a robust web interface with related GIS data for evaluating results.