INFORMATION TECHNOLOGY FOR DETECTING HEAT CHANGES IN URBAN AREAS BASED ON MACHINE LEARNING AND SATELLITE DATA
DOI:
https://doi.org/10.35546/kntu2078-4481.2026.2.31Keywords:
machine learning, Isolation Forest, Random Forest, anomalies, heat changes, urban heat islands, satellite dataAbstract
The paper presents an information technology for detecting heat changes in urbanized areas by integrating satellite data with machine learning methods. The proposed approach combines nonlinear regression modeling of Land Surface Temperature using the Random Forest algorithm with unsupervised anomaly detection based on the Isolation Forest algorithm. Multispectral imagery from Landsat 8, Landsat 9, and Sentinel-2 missions for the period 2020–2024 was used to construct a multidimensional feature space including spectral indices, texture measures, and morphometric parameters. This approach is particularly relevant for areas undergoing significant anthropogenic and natural changes. An example of this is the territory of the city of Nikopol and the adjacent water area of the former Kakhovka Reservoir, which underwent significant changes after the destruction of the Kakhovka HPP dam in 2023. The disappearance of the water surface transformed the cooling water bodies into dry, open soil, affecting surface reflectivity and heat accumulation. It created conditions for abnormal overheating and intensified the urban heat island effect, which was confirmed by experimental studies. Quantitative metrics showed the high accuracy of the proposed approach, namely, RMSE = 2.12 °C, MAE = 1.63 °C, and R² = 0.89. A substantial increase in the proportion of areas characterized by anomalous temperature values was identified, rising from 15–30% in 2020–2021 to 85–90% in 2024. It indicates a transition from a spatially localized overheating pattern to a regionally extensive thermal field. The scientific novelty consists in improving the approach to determining thermal changes in urbanized areas based on satellite data, which, unlike existing methods that are limited either to regression modeling of the Earth's surface temperature or to separate classification of urban heat island zones, involves the use of ensemble nonlinear regression determination of anomalies with an unsupervised Isolation Forest algorithm for automated detection and mapping of anomalous temperature areas
References
Wang W., Yao X., Shu J. Air advection induced differences between canopy and surface heat islands. Sci. Total Environ. 2020. Vol. 725. 138120.
Sobrino J. A., Irakulis I. A Methodology for Comparing the Surface Urban Heat Island in Selected Urban Agglomerations Around the World from Sentinel-3 SLSTR Data. Remote Sens. 2020. Vol. 12. 2052.
Hemmatian F., Sohrabi M. K. A survey on classification techniques for opinion mining and sentiment analysis. Artificial Intelligence Review. 2019. Vol. 52, No. 3. P. 1495–1545. DOI: 10.1007/s10462-017-9599-6.
Meng Q., Zhang L., Sun Z., Meng F., Wang L., Sun Y. Characterizing spatial and temporal trends of surface urban heat island effect in an urban main built-up area: A 12-year case study in Beijing, China. Remote Sens. Environ. 2018. Vol. 204. P. 826–837.
Каштан В. Ю., Гнатушенко В. В. Інформаційна технологія виявлення контурів лісових пожеж за оптичними супутниковими даними. Системні технології. Регіональний міжвузівський збірник наукових праць. 2023. Вип. 1 (144). С. 3–12. DOI: 10.34185/1562-9945-1-144-2023-01.
Liu X., Ming Y., Liu Y., Yue W., Han G. Influences of landform and urban form factors on urban heat island: Comparative case study between chengdu and chongqing. Sci. Total Environ. 2022. Vol. 820 (5). 153395. DOI: 10.1016/j. scitotenv.2022.153395.
Каштан В. Ю., Сергєєва К. Л., Коробко О. В., Іванов Д. В. Пошук та оцінка островів тепла на цифрових космічних знімках. Системні технології. Регіональний міжвузівський збірник наукових праць. Дніпро, 2023. Вип. 3 (146). С. 87–98. DOI: 10.34185/1562-9945-3-146-2023-09.
Deilami K., Kamruzzaman M., Liu Y. Urban heat island effect: A systematic review of spatio-temporal factors, data, methods, and mitigation measures. Int. J. Appl. Earth Observation Geoinformation. 2018. Vol. 67 (5). P. 30–42. DOI: 10.1016/j.jag.2017.12.009.
Yu X., Guo X., Wu Z. Land Surface Temperature Retrieval from Landsat 8 TIRS – Comparison between Radiative Transfer Equation-Based Method, Split Window Algorithm and Single Channel Method. Remote Sens. 2014. Vol. 6. P. 9829–9852.
Zhou L., Yuan B., Hu F., Wei C., Dang X., Sun D. Understanding the effects of 2D/3D urban morphology on land surface temperature based on local climate zones. Build. Environ. 2022. Vol. 208 (1). 108578. DOI: 10.1016/j. buildenv.2021.108578.
Tanoori G., Soltani A., Modiri A. Machine learning for urban heat island (UHI) analysis: Predicting land surface temperature (LST) in urban environments. Urban Clim. 2024. Vol. 55. 101962.
Adeniran I. A., Nazeer M., Wong M. S., Chan P. W. An improved machine learning-based model for prediction of diurnal and spatially continuous near surface air temperature. Sci. Rep. 2024. Vol. 14. 27342.
Каштан В. Ю., Шевцова О. С. Інформаційна технологія попередньої обробки супутникових зображень з використанням згорткової нейронної мережі. Системні технології. Регіональний міжвузівський збірник наукових робіт. Дніпро, 2024. Вип. 1 (150). С. 36–50. DOI: 10.34185/1562-9945-1-150-2024-04.
Isa M. S., Suharjito S., Kusuma G., Cenggoro T. W. Supervised conversion from Landsat-8 images to Sentinel-2 images with deep learning. European Journal of Remote Sensing. 2021. Vol. 54. P. 182–208. DOI: 10.1080/22797254.2021.1875267.
Mwangi P., Karanja F., Kamau P., Letema S. Contribution index of land cover and land surface temperature changes in upper hill Nairobi, Kenya. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences. 2021. Vol. V-3-2021. P. 141–149. DOI: 10.5194/isprs-annals-V-3-2021-141-2021.
