High-Resolution Model Accuracy During Extreme Heat Events on Neighborhood Scales: A Case Study in Baltimore, Maryland

With the changing climate, there is a heavy focus on the future climate and how we will adapt our models to accurately represent the quickly changing scenarios. Of interest is model performance and depiction of the heterogeneity of meteorological conditions at neighborhood scales in urban environments. More specifically, the performance of high-resolution models during extreme weather events, like heat waves. In partnership with the Baltimore Social-Environmental Collaborative (BSEC) project, the High-Resolution Rapid Refresh (HRRR) model was evaluated against the observations collected by BSEC. Temperature and relative humidity data were analyzed from 28 stations across Baltimore, Maryland, between June 16 and June 22, 2024. Observations were matched to grid points in the HRRR 00-hour forecast. After significance testing, it was found that the HRRR underestimates the urban heat island in Baltimore, Maryland, by 0.5 – 1.5oC in urban areas during the overnight hours. With regard to relative humidity, the HRRR underestimated values by 5% – 15%. Potential causes of this underestimation are small-scale surface fluxes and poor radiative transfer in urban areas that are not represented accurately in the model dynamics.