Xuewei Wang, Angel Hsu, TC Chakraborty 

Through conversion of land cover to more built-up, impervious surfaces, cities are creating hotter environments for urban residents. Existing measurements of heat and heat stress, however, are often insufficient to capture intra-urban variability of exposure. This study provides a replicable method for modeling air temperature, humidity, and heat stress over an urban area, engaging citizens in collecting high-temporal and spatially-resolved air temperature and humidity measurements. We use low-cost, consumer-grade sensors combined with satellite remote sensing data and machine learning to map urban air temperature and humidity over various land-cover classes to understand intra-urban spatial variability of heat at a relatively high resolution (10 meters). Our findings show that individuals may be exposed to higher levels of heat and heat stress than weather station data suggest, and this heat varies according to land cover type, with tree-covered land the coolest and built-up areas the warmest, and time of day, with higher temperatures observed during the early afternoon. Combining our resulting dataset with sociodemographic data, we find an inverse relationship between income and our heat metrics, although the sensitivity of this relationship varies depending on which metric is used. Policymakers and urban planners can use this data to identify areas exposed to high heat and heat stress as a first step to design effective mitigation measures.