基于高分卫星数据多时相重建的水体信息提取  

Extraction of water body information from GF-1 satellite images using spatiotemporal reconstruction approach

作　　者：白亮亮 曾超[2] 盖长松[3] 黄琦 龙笛[1] BAI Liangliang;ZENG Chao;GE Changsong;HUANG Qi;LONG Di(State Key Laboratory of Hydroscience and Engineering,Tsinghua University,Beijing 100084;School of Resource and Environmental Sciences,Wuhan University,Wuhan 430079;Chongqing Meteorological Service,Chongqing 401147)

机构地区：[1]清华大学水沙科学与水利水电工程国家重点实验室,北京100084 [2]武汉大学资源与环境科学学院,武汉430079 [3]重庆市气象局,重庆401147

出　　处：《水力发电学报》2021年第2期111-120,共10页Journal of Hydroelectric Engineering

基　　金：国家重点研发计划(2017YFC0405802,2017YFC0405801);重庆市气象局开放式研究基金项目(KFJJ-201502)。

摘　　要：基于我国高分一号卫星WFV数据,利用其较高的时空分辨率特性,建立库区高分影像时间序列与水体“检测‒重建‒提取”的流程框架,以提高面向三峡库区的水体分布及面积变化的动态监测能力。研究结果表明:基于多特征联合的面向对象的云检测总体精度高达96.8%,且云阴影的生产者与用户精度均高于76%。多时相重建算法可以对云覆盖区域进行有效的填补,且重建后的影像可反映出完整的空间变化和差异性。从2013—2017年的水体面积提取结果可知,三峡库区水体面积并未出现明显增大或减小的趋势,多年平均水体面积为540.6 km2。研究结果为三峡库区综合治理提供了数据和技术支撑,提高了高分卫星数据在三峡库区水体面积提取中的应用价值和能力。To facilitate dynamic monitoring of the spatial distribution of water bodies in the Three Gorges reservoir area,we have developed a framework of water body detection,reconstruction and extraction using China’s GF-1 satellite images of high spatiotemporal resolution.Results show the method of objectoriented change detection has an overall accuracy of cloud detection up to 97%,with the producer's accuracy and user's accuracy of cloud detection both higher than 76%.Large cloud-contaminated areas can be effectively recovered using a spatiotemporal reconstruction approach,which can reflect the complete spatial variations.In addition,we find the water surface area of this reservoir,with a mean of 540.6 km2,did not significantly increase or decrease during 2013-2017.This study is useful for comprehensive,integrated management of the reservoir area,with important implications for applying GF-1 satellite images in this region.

关 键 词：三峡库区 高分一号 云检测 时空重建 水体面积变化 

分 类 号：P343[天文地球—水文科学]