DETECTION AND CHARACTERIZATION OF URBAN HEAT ISLANDS WITH MACHINE LEARNING

Abstract

Assessing and understanding the urban scale impacts of extreme climate events is a global necessity. Risks associated with heat, where intra-urban dynamics and rural/urban boundary conditions greatly impact its distribution, is of particular interest as the evolution of climate change and urbanization persists. Characterizing Urban Heat Island (UHI) effects is dependent on the availability of, among other, high-resolution, near-surface air temperature maps as well as a description of the Local Climate Zones (LCZ). This study assesses the applicability of state-of-the-art (SOTA) Artificial Intelligence (AI) techniques for UHI detection and characterization. We fine-tune a Geospatial Foundation Model (GFM) to predict 2m air temperature at a 1km resolution for the urban areas of Johannesburg, South Africa, with average absolute error measures less than 1.5°C, and further enabled UHI characterization through a pixel-based AI model for LCZ classification for the same region of interest.