Recent Improvements in Supervised Pixel-Based LCZ Classification

Accurate local climate zone (LCZ) maps are crucial for urban environmental studies. The last chapter introduced the ways to generate LCZ maps, including object-based and pixel-based remote sensing and GIS methods. Among these methodologies, the supervised pixel-based method using open-access remote sensing imagery has gained popularity, providing a fast and cost-efficient way for LCZ classification. Implementing the World Urban Database and Access Portal Tools (WUDAPT) further provides an open platform and global database for consistent supervised pixel-based LCZ information to support different types of applications and research ( http://www.wudapt.org/ ). This chapter outlines three critical components in supervised pixel-based LCZ classification, including (1) geometrical pre-processing and classification platform (Sect. 4.1), (2) remote sensing data (Sect. 4.2); and (3) classification algorithm (Sect. 4.3), and their recent development and improvement in LCZ classification. Three case studies comparing different classification algorithms in Asian cities’ LCZ mapping (Sects. 4.4, 4.5, and 4.6) are also presented in this chapter.