Adviser: Ken Davis

Heat and cold stress are the leading causes of weather-related deaths, and the physical properties of cities can exacerbate the impacts of heat and cold stress. Therefore, it is important to understand and accurately model these hazards at fine spatial scales so that mitigation strategies can be developed. The goals of this work are to assess a weather modeling system’s ability to accurately represent heat and cold stress, determine if the modeling system contains biases, and explore sources of systematic errors. In this study, output from a five-year, 1km2 resolution Weather Research and Forecasting Model (WRF) simulation of the Indianapolis, Indiana climate is compared to a high-resolution network of observations. Results show that the model heat index is most accurate during summer days as the domain-wide biases average 0.18°C, but the magnitude of the biases are suppressed by compensating errors of the temperature and relative humidity fields. The modeling system fails to accurately capture nighttime heat stress as the neighborhood-level urban biases range from -0.13°C to 4.5°C. Analysis of surface fluxes indicates that urban fraction errors drive large biases in sensible and latent heat fluxes, leading the model to have a too hot and dry urban atmosphere. Wintertime cold stress is not captured well as the domain-wide windchill biases are -4.8°C and -7.09°C during the daytime and nighttime respectively. Overall, these results imply that updates to the landcover representation can alter fluxes of heat, moisture, and momentum, and this can contribute to more accurate fine-scale heat stress predictions within Indianapolis.

Zoom link: https://psu.zoom.us/j/92611172572