本文探讨了使用Sentinel-1双极化数据进行全球范围内本地气候区域分类的挑战，特别是在不同城市间知识转移的困难。研究发现，随机森林算法在不同城市间直接应用时，准确率大幅下降，强调了模型泛化能力的重要性。文章介绍了SentinelSat和SNAP两个开源工具，分别用于搜索、下载Sentinel卫星图像及其元数据。
Feature Extraction and Selection of Sentinel-1 Dual-Pol Data for Global-Scale Local Climate Zone Classification
However, this is a challenging task. For example, one study [20] attempted to
select training samples from one city for the classification of another city using RF. The classification accuracies dropped to 18.2%, which indicates that the knowledge transferability between different cities should be carefully considered.
develop a classifier with adequate generalization capability to be applied to any other cities
An open-source toolbox, named SentinelSat (https://github.com/sentinelsat/sentinelsat), provides the utilities of searching, downloading, and retrieving the metadata of Sentinel satellite images.
Sentinel Application Platform (SNAP, https://step.esa.int/main/toolboxes/snap/)
扫一扫
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
