摘要
以中国甘肃省张掖地区为研究区,获取了2018年3月—10月间该地区的晴天GF-4卫星数据、7月30日的哨兵二号卫星数据以及该地区的SRTM DEM(digital elevation model)数据,研究探讨基于多源数据的全特征土地分类方法。结果表明,全特征的多源数据能够较好地实现土地覆盖分类,本研究区的分类精度达90.22%,并具有较好的可信度。所提出的多源数据的全特征土地分类方法为后续高分系列卫星的综合应用提供了重要的参考价值。
The study takes Zhangye area of Gansu Province in China as an example,and obtaines GF-4 satellite data from March to October 2018,the Sentinel II satellite data of July30,and the SRTM DEM data of this area.The full-feature land cover classification method is developed based on the spacial,temporal,and spectral features extracted from these multi-source data.The results show that the multi-source data could provide full features of land cover classification,the total classification accuracy of which reaches 90.22%and has good credibility,.The full-feature land cover classification method proposed in this study provides an important reference for the comprehensive utilization of subsequent high-resolution satellite series.
作者
张煜辉
ZHANG Yuhui(Wuda Geoinformatics Co,Ltd.,Wuhan 430223,China)
出处
《测绘地理信息》
CSCD
2022年第4期90-94,共5页
Journal of Geomatics
基金
国家重点研发计划(2016YFB0500304)
关键词
高分四号
多源数据
土地覆盖分类
全特征
GF-4
multi-source data
land cover classification
full features
