Comparing Object-Based and Pixel-Based Methods for Local Climate Zones Mapping with Multi-Source Data

The local climate zones (LCZs) system, a standard framework characterizing urban form and environment, effectively promotes remote sensing research, especially heat island (UHI) research. However, whether mapping with objects is more advantageous than pixels in LCZ remains uncertain. This study aims to compare object-based pixel-based multi-source data detail. By comparing the method 50 100 m, respectively, we found that performed better overall accuracy (OA) higher at approximately 2% 5%, respectively. In per-class analysis, showed clear advantage land cover types competitive performance built while LCZ2, LCZ5, LCZ6 m. We further employed correlation-based feature selection (CFS) evaluate importance paradigm, finding building height (BH), sky view factor (SVF), surface fraction (BSF), permeable (PSF), use exhibited high frequency image bands were scarcely selected. summary, concluded capable of performs under same training condition unless under-segmentation cases.