Abstract
Abstract
Cooling energy demand is sensitive to urban form and socioeconomic characteristics of cities. Climate change will impact how these characteristics influence cooling demand. We use random forest machine learning methods to analyze the sensitivity of cooling demand in Chicago, IL, to weather, vegetation, building type, socioeconomic, and control variables by dividing census tracts of the city into four groups: below-Q1 income–hot days; above-Q1 income–hot days; below-Q1 income–regular days; and above-Q1 income–regular days. Below-Q1 census tracts experienced an increase in cooling demand on hot days while above-Q1 census tracts did not see an increase in demand. Weather (i.e. heat index and wind speed) and control variables (i.e. month of year, holidays and weekends) unsurprisingly had the most influence on cooling demand. Among the variables of interest, vegetation was associated with reduced cooling demand for below-Q1 income on hot days and increased cooling demand for below-Q1 income on regular days. In above-Q1 income census tracts building type was the most closely associated non-weather or control variable with cooling demand. The sensitivity of cooling demand for below-Q1 income census tracts to vegetation on hot days suggests vegetation could become more important for keeping cities cool for low-income populations as global temperatures increase. This result further highlights the importance of considering environmental justice in urban design.