Estimating the anthropogenic contribution to the summer urban heat island in a Mediterranean coastal city - the case of Tel Aviv, Israel

In areas with a long summer season, especially where heat stress prevails throughout the day, such as Mediterranean coastal cities, Urban Heat Island (UHI) causes further thermal discomfort and an increase in energy consumption. As the climate continues to warm, additional cooling of homes and public spaces will likely be required, creating a feedback loop and raising temperatures even further. Previous studies have identified a pronounced UHI in Tel Aviv under favorable synoptic conditions, but none have attempted to quantify the direct contribution of urban waste heat from buildings and transportation to the UHI. This study analyses the characteristics of the summer UHI of Tel Aviv under different weather conditions, using meteorological data combined with energy consumption data from two seasons (2023-2024), and develops a statistical downscaling model that predicts the UHI intensity (UHII), based on synoptic, mesoscale and anthropogenic thermal forcing variables. A refined model assessing the direct contribution of waste heat from anthropogenic emissions based on synoptic conditions was developed and exhibited a logarithmic correlation of R=0.83 between the calculated anthropogenic forcing and the UHII.  While energy consumption, primarily from the use of air conditioning (AC) was strongly correlated with nocturnal heat stress, a conflicting finding was the inverse relationship between the nocturnal temperature and the UHII.   A multi-regression statistical predictive model for the UHI shows that while anthropogenic forcing is positively correlated to the UHII, and its direct contribution can be estimated at 0.5°C-1°C during most nights, the direct anthropogenic contribution is overwhelmingly determined by synoptic conditions and not by variations in heat flux. Further research in cities with similar climates could provide recommendations for improved mitigation of waste heat dispersal thus limiting the predicted feedback.