High-resolution mapping of urban wet-bulb temperature and its intra-urban variability: A London case study

Most extreme-heat studies map high-resolution near-surface dry-bulb air temperature ( T a ), yet wet-bulb temperature ( T w ) fields that can capture the combined effects of T a and humidity to represent human heat stress remain limited. This data gap constrains investigation of intra-urban heat variability and the timing of peak heat stress. Here, based on the Rankine-Kirchhoff approximations of Romps, this study innovatively generates hourly 1-km T w by integrating crowdsourced station observations with remote sensing and reanalysis data in machine learning based downscaling frameworks. Taking London during the summers of 2019–2022 as a case study, the tuned RF-based downscaling approach achieved the best performance ( MAE = 0.65 °C, RMSE = 0.93 °C, R 2 = 0.91) among all models. The resulting T w patterns diverge from those commonly reported for T a : blue infrastructure shows no consistent cooling effect, whereas vegetated areas cool T w increasingly as tree cover rises, pointing to tree-dominated green infrastructure as a promising humid-heat countermeasure. Results further show that daily maximum T w occurs 0.7 h earlier than daily maximum T a in London, opposite to findings elsewhere, implying that climate type and coastal influence modulate the timing of peak T w , which may affect the attribution of excess nighttime mortality. Besides, a systematic global assessment of the mechanisms controlling T w extremes is also required in the future.