Combination of Sentinel-2 and PALSAR-2 for Local Climate Zone Classification: A Case Study of Nanchang, China

Local climate zone (LCZ) maps have been used widely to study urban structures and heat islands. Because remote sensing data enable automated LCZ mapping on a large scale, there is need evaluate how well resources can produce fine assess thermal environments. In this study, we combined Sentinel-2 multispectral imagery dual-polarized (HH + HV) PALSAR-2 generate of Nanchang, China using random forest classifier grid-cell-based method. We then the importance scores different input features (Sentinel-2 bands, channels, textural features) for classification model their contribution each class. Finally, investigated relationship between LCZs land surface temperatures (LSTs) derived from summer nighttime ASTER by spatial statistical analysis. The highest accuracy was 89.96% when all were used, which highlighted potential data. most important feature short-wave infrared-2 band Sentinel-2. spectral reflectance more than polarimetric in classification. beneficial several cover types combined. Summer LSTs differed significantly other. Results also demonstrated that grid-cell processing provided homogeneous usual resampling methods. This promising reference further improve quantitative analysis local climate.