UHI is an important measure for understanding the urban landscape, especially in terms of thermal agglomeration and disturbance. This research aims to discern the success of sustainability planning by examining and comparing the different characteristics of UHIs through the combination of machine learning and statistical methods. To achieve this, we analyze 4 new towns in Korea, which include two ‘old’ new towns and two ‘recent’ new towns. The key difference between our test towns lies on whether or not the sustainability policies were applied to their development plans. We visualize LST and conduct a k-mean clustering to find and quantify spatial patterning in the resulting UHI measures. We then compare the statistical relations between LST and 6 UHI driven variables across the towns. Using comparative analysis, this research reveals that sustainable development policies have a notable effect on the patterns and intensities of UHI. Urban structures, planned under development policies, including green and blue space ratios, road networks, and housing distributions, were found to affect UHI significantly. We quantifiably confirm that the sustainability policies implemented in planning the ‘recent’ new towns allow the towns to experience less aggravated UHIs than the ‘old’ new towns. However, we also claim a need to develop appropriate, long-term UHI management regulations for the ‘recent’ new towns. This paper provides a solid basis for improving Korean new town planning and managing the environmental issues in urban systems for planners, designers, and decision-makers to establish the sustainable built environment.

UHI是了解城市景观的一项重要措施，尤其是在热集聚和扰动方面。这项研究旨在通过结合机器学习和统计方法来检查和比较UHI的不同特征，从而识别可持续性计划的成功。为此，我们分析了韩国的4个新城镇，其中包括两个“旧”新城镇和两个“最近”新城镇。我们的测试城镇之间的主要区别在于是否将可持续性政策应用于其发展计划。我们想象LST并进行ķ-均值聚类，以发现和量化所得UHI度量中的空间模式。然后，我们比较了各个城镇的LST和6个UHI驱动变量之间的统计关系。通过比较分析，该研究表明，可持续发展政策对UHI的模式和强度具有显着影响。人们发现，根据发展政策规划的城市结构，包括绿色和蓝色空间比例，道路网络和住房分布，对UHI产生了重大影响。我们可以量化地确认，在规划“最近的”新镇时实施的可持续性政策使城镇经历的“重症监护”比“旧的”新镇少。但是，我们还声称有必要为“最近的”新城镇制定适当的长期UHI管理法规。