Using classification algorithms to model nighttime Surface Urban Heat Island (SUHI), with an emphasis on the role of urban trees

[EN] The Urban Heat Island (UHI) effect represents the increased temperature in urban areas compared to their surroundings. This is often addressed using Land Surface Temperature (LST), thereby accounting for the so-called Surface UHI (SUHI). Few studies have incorporated urban trees into such models as a starting point for analyzing the influence of their characteristics on temperature reduction. In response, we used classification algorithms to model SUHI as a binary variable from values of nighttime LST, focusing on the cooling effect of urban trees. The results of a case study in the city of Valencia (Spain) demonstrated the high accuracy of Support Vector Machines (SVM) in classification, achieving Area Under the Curve (AUC) values of over 80 %. Urban trees were one of the most relevant variables for predicting SUHI due to their ability to offset part of the warming effects of buildings. Trees such as Melia azedarach, Pittosporum tobira and Ulmus minor were most effective due to their physiognomy and morphology. Therefore, climate resilience-oriented planning in Valencia should prioritize planting tree species with such characteristics to better regulate temperature.