Exploring Urban Neighborhood-Scale Heat Warning Criteria Based on Temporal and Spatial Threshold Exceedances

Urban heat stress is an escalating challenge as climate change amplifies the frequency and intensity of heat waves worldwide. This requires appropriate actions at the local-scale, including early warning systems to prepare locally for heat actions and improve societal resilience in cities. However, existing heat warning systems often lack sufficient spatial resolution to account for the significant urban-rural and intraurban variabilities, explained in particular by delayed night-time cooling and daytime overheating of impervious open areas. This study proposes a framework for defining neighborhood-scale heat warning criteria, leveraging threshold exceedances of the Universal Thermal Climate Index (UTCI) derived from high-resolution geospatial and model data.

Our work utilizes outputs from the Human Thermal Comfort Neural Network (HTC-NN), a machine-learning-based emulator of numerical urban climate models. The HTC-NN allows for rapid predictions of key atmospheric variables which are synthesized to calculate pedestrian-level UTCI values at high spatial (1 x 1 m) and temporal resolution (1 hr) based on regional-scale weather forecasts.

In this study, HTC-NN predictions for summer 2023 were run for Freiburg, Germany forced with ERA5 Land data. We aggregate 1 x 1 m predictions of UTCI to 500 x 500 m heat distribution statistics and calculate exceedance fractions of relevant UTCI heat stress classes at the neighbourhood scale. A spatio-temporal analysis is conducted separately for night and day to investigate the relationship between exceedance fractions, meteorological conditions, and land cover fractions exploring potential neighbourhood heat warning criteria and thresholds. Results reveal pronounced intra-urban variability and strong diurnal variations of exceedance fractions, emphasizing the interplay of urban form, shading effects, and surface properties. These findings highlight the need to differentiate between urban and rural areas and also between different urban environments in terms of heat stress warnings.