Predictive analysis of Urban Heat Islands using satellite data and neural network algorithms

The Urban Heat Island (UHI) phenomenon is a critical concern, particularly in the context of global warming and rapid urbanization. UHIs are essentially urbanized areas that exhibit higher temperatures compared to their less or non-urbanized surroundings. This heat island effect is worsened by urbanization, largely due to the extensive use of asphalt and other impervious surfaces over green spaces, coupled with various human activities. The environmental conditions created by UHIs negatively impact the quality of life. These areas suffer from elevated temperatures, higher concentrations of pollutants, and a subsequent increase in the energy and economic costs associated with cooling buildings. Numerous studies have been carried out to tackle the growing issue of the UHI. These efforts concentrate on analyzing UHI features to equip environmental planners and decision-makers with vital instruments for mitigation and management. This work investigates the UHI phenomenon in the Catania area (Sicily, Italy), focusing on a specific urban section to highlight the contrast between densely built and greener spaces. The study employs remote sensing data from Landsat 8 and 9 satellite missions to calculate relevant indices, such as the Normalized Difference Vegetation Index (NDVI) and Land Surface Temperature (LST), which are essential for UHI analysis. After generating a thematic map of UHIs for the area, the Land Use Land Cover (LULC) was analyzed. This LULC analysis facilitated the use of the QGIS MOLUSCE plug-in, a tool offering several algorithms for predictive LULC modeling. The available algorithms include neural networks (multilayer perceptron), logistic regression, weights of evidence, multi-criteria evaluation, and validation via kappa statistics. The model's results were validated by projecting them onto a year for which actual data was already available. Predictive LULC modeling enables the evaluation of UHI conditions at the time of the projection. This capability makes the tool valuable for environmental planners and decision-makers, aiding in the assessment of future urbanization impacts and their subsequent effects on the population's quality of life.