基于谷歌地球引擎的开封城区2010—2019年水体分布变化研究

为了研究开封市年际水体分布变化,选用Landsat系列多时相影像数据,在谷歌地球引擎google earth engine(GEE)云平台上结合归一化水指数、改进的归一化水指数、归一化植被指数以及归一化建筑指数构建光谱特征使用分类回归树、支持向量机、随机森林分类方法对开封市城区的水体进行提取并分析.结果表明:基于GEE云平台和机器学习的分类,能够很好地提取出开封市城区年际水体分布变化情况;开封市城区的水体面积处于不断变化中,先减少后增加,总体上趋于增加,随着城市的发展变化,城区的景观格局发生较大改变,影响开封市水体面积产生变化的主要因素是北部黄河水量的变化以及开封市近年来正在实施的"城市双修"对水体的影响;使用随机森林分类能更好地对开封市城区的水体进行提取,提取水体的最小精度为94.6％,平均总体分类精度为96.9％,平均Kappa系数为0.954.

Study on Water body change in Kaifeng Urban from 2010 to 2019 based on Google Earth Engine

In order to study the interannual change of water body in Kaifeng city, Landsat series multi-temporal image data were selected, and the spectral features were constructed by combining normalized water index, improved normalized water index, normalized vegetation index and normalized building index on Google Earth Engine (GEE) cloud platform. Classification regression tree, support vector machine and random forest classification were used to extract and analyze the water body of Kaifeng city. The results show that the interannual changes of water bodies in Kaifeng city can be well extracted based on the classification of GEE cloud platform and machine learning. The water body area of Kaifeng city is constantly changing, decreasing at first and then increasing, and tends to increase on the whole. With the development and change of the city, the landscape pattern of the city has changed greatly. The main factors affecting the change of water body area in Kaifeng city are the change of the water volume of the Yellow River in the north and the influence of the " Urban double repair" being implemented in Kaifeng city in recent years. The use of random forest classification can better extract the water body of Kaifeng city, the minimum accuracy of water extraction is 94.6%, the average overall classification accuracy is 96.9%, and the average kappa coefficient is 0.954.