Automatic mapping of urban heat islands based on high-resolution climate simulations with application to French cities

Source abstract

Natural or rural areas are characterised by minimal or absent urbanisation. They serve as crucial benchmarks for impartial comparisons with cities. Urbanisation raises both surface and air temperatures compared to rural areas, creating the urban heat island (UHI) effect. A major challenge in assessing UHI is finding suitable rural reference areas for each city. This study presents a method that uses high-resolution climate simulations from the regional climate model CNRM-AROME cy46t1 (2.5 km resolution) to systematically identify reference rural areas and measure UHI intensity at the grid-cell level. The model integrates the classification of the ECOCLIMAP land cover, derived from high-resolution CORINE Land Cover data, to extract urban areas without the need for prior regional knowledge or defined climate zones. The developed method, called MOD\_Mask, was tested across metropolitan France, including cities of varying sizes, from the densely populated Paris region to the medium-sized city of Dijon. The data from the local weather station helped evaluate the automated UHI mapping method in these two cities. MOD\_Mask successfully identified and mapped the areas most exposed to UHI effects with high precision and helped investigate population exposure in 12 French cities. The methodology holds significant potential for future climate projections and scenario analyses, offering critical insights into key climate indicators. It provides spatial diagnostics that help decision-makers understand UHI conditions and shape climate adaptation strategies. The framework can also be applied to other regions and expanded to include more climate variables.