Satellite and ground-based data to monitor urban heat islands. Cases of study: polish and italian cities
Keywords: Thermal sensors, Landsat satellite images, Land Surface Temperature, Urban Heat Islands, Climate change, Landsat
Abstract. Urban Heat Island (UHI) is a significant environmental phenomenon that exacerbates rising temperatures in urban areas, affecting human health, energy consumption, and urban sustainability. This study analyzes UHI intensity in four selected cities—Kraków and Gdańsk (Poland) and Ancona and Termoli (Italy)—using Landsat-8 and Landsat-9 thermal infrared imagery combined with ground-based in situ temperature measurements. The study employs object-based classification using a Random Forest (RF) algorithm, enabling a detailed assessment of land surface temperature (LST) variations across different land cover types.
The results indicate that built-up areas and impervious surfaces exhibit significantly higher temperatures, with LST values reaching 36.5°C in Ancona and 32.2°C in Kraków, whereas vegetated areas show a cooling effect, reducing temperatures by up to 7°C. Coastal cities, particularly Gdańsk and Termoli, display lower UHI intensity but greater LST variability due to maritime cooling and cloud cover effects.
Statistical validation comparing satellite-derived LST with in situ measurements demonstrated a strong correlation in Kraków (r = 0.73) and moderate agreement in Ancona (r = 0.62), while Gdańsk and Termoli exhibited lower correlation values (r = 0.09 and r = 0.003, respectively), attributed to cloud interference and coastal microclimatic effects. The study highlights the importance of integrating remote sensing with ground-based data to enhance UHI monitoring accuracy and proposes urban planning strategies, including increased vegetation cover and heat-resilient urban materials, to mitigate UHI effects.
These findings contribute to the scientific understanding of UHI dynamics in different climatic contexts and provide actionable insights for sustainable urban development. Future research should explore high-resolution thermal datasets and advanced machine learning models for improved urban temperature monitoring.