Urban areas are increasingly vulnerable to the impacts of climate change, including elevated temperatures caused by the urban heat island effect. This study evaluated the effectiveness of Nature-Based Solutions (NBSs) in mitigating urban surface temperatures in Guimarães, Portugal, a city recognized for its sustainable urban planning initiatives. Integrating remote sensing data, machine learning models, and spatiotemporal analysis, it examined the evolution of land use and land cover (LULC) and land surface temperature (LST) from 2013 to 2023, with predictions for 2028.
This study employed Landsat 8 imagery to derive LST, NDVI, and NDBI indices and classified LULC using the Random Forest algorithm. A progressive increase in vegetated areas from 154.76 km² in 2013 to 172.22 km² in 2023 reflects the influence of urban sustainability initiatives such as the "Guimarães Mais Floresta" program. Machine learning models, including XGBoost, Bagging, and AdaBoost, were used to predict LST for 2028. XGBoost outperformed others, achieving an R² of 0.9543 and providing precise predictions for urban planners.
The results highlight that NBS implementations, such as green roofs and urban gardens, reduced local temperatures by up to 2.49 °C. However, projections for 2028 indicate a slight reduction in vegetated areas, underscoring the need for stronger environmental policies. This study also identified thermal hotspots, predominantly in built-up areas, where temperatures are expected to exceed 37 °C in 2028. These findings emphasize the importance of targeted NBS interventions in urban planning to mitigate climate risks.
This research advances spatiotemporal methodologies by combining multitemporal remote sensing data, machine learning predictions, and local validation. The approach offers a replicable framework for assessing the effectiveness of NBSs and provides recommendations for sustainable urban development. Future studies could expand the temporal dataset with additional satellite imagery and test this methodology in cities with diverse climatic and urbanization patterns, offering broader insights into the effectiveness of NBSs.
