The local climate zone (LCZ) scheme efficiently associates urban forms with thermal environments. However, confronted with diversified urban fabrics, the initial LCZ scheme has encountered challenges of its restricted classes and intraclass temperature differences. This study aimed to develop an expanded LCZ scheme to accommodate the diversified urban morphological evolution. Thirty-two block prototypes were developed according to the building footprint ratios (BFRs), mean building heights (BHs), and mean sky view factors (SVFs) of the blocks in Shanghai, China. Then, the block prototypes were classified based on their morphological and temperature indices by k-means clustering, an unsupervised clustering method that maximizes the distance between the clusters and minimizes the distance within the cluster. Results indicated that the block prototypes were clustered into 14 classes, the high-rise LCZs were subdivided, and blocks within six classes were not covered by the initial LCZ scheme. Accordingly, an expanded LCZ scheme was developed by introducing four novel LCZ types and adjusting and subdividing the initial LCZ types, which covered 35% more blocks than the initial LCZ scheme. The mean SVF demonstrated the most significant correlation with the temperature of LCZs, whereas the effects of BFR and BH on the temperature depended on LCZ types.