Modelling Urban Heat Island with Landsat Images and LiDAR Datasets

Abstract

The environmental problems caused by rapid and uncontrolled urbanization are increasing day by day. When the needs of the growing population are also in question, urbanization problems become unavoidable when considered on a global scale. Since housing is one of the most important population-related needs, there is an increase in urban areas and an expansion of working areas. Therefore, while urban areas are increasing, forested or green areas are shrinking. Decreasing green areas play a very important role in increasing surface temperature. This increase in cities, which are warmer than the surrounding rural areas, is briefly called Urban Heat Island (UHI). Urban Thermal Field Variance Index (UTFVI) was used to calculate the UHI effect. This index, produced from thermal tapes, determines the severity of UHI. For this study, attributes were created using Landsat, LiDAR, and Open Street Map data to estimate UHI severity without the aid of thermal bands. Using these features, a random forest machine learning algorithm was trained, and UHI severity maps were estimated. For the city of Sioux Falls, which was selected as the study area, data for the years 2008 and 2020 were used. UHI forecast maps for these years were produced and compared with ground truth maps.