The local climate zones classification takes input from one Sentinel-1 image and four seasonal Sentinel-2 images as inputs. The Sentinel-1 images were downloaded from ESA SciHub, and prepared using ESA SNAP software. Sentinel-2 images were semi-automatically downloaded and prepared using Google Earth Engine and MATLAB. The local climate zones classification labels were predicted using convolutional neural network with a model pre-trained on the "So2Sat LCZ42" training data (https://doi.org/10.14459/2018MP1454690). A demo classification script with the trained model can be found on https://github.com/zhu-xlab/So2Sat-LCZ-Classification-Demo.    «

The local climate zones classification takes input from one Sentinel-1 image and four seasonal Sentinel-2 images as inputs. The Sentinel-1 images were downloaded from ESA SciHub, and prepared using ESA SNAP software. Sentinel-2 images were semi-automatically downloaded and prepared using Google Earth Engine and MATLAB. The local climate zones classification labels were predicted using convolutional neural network with a model pre-trained on the "So2Sat LCZ42" training data (https://doi.org/10.14...    »