基于哨兵2号的复杂地域水体提取研究  

Research on Water Body Extraction of Complex Area Based on Sentinel-2

作　　者：任源鑫 林青 李明洁 周旗[2,3] REN Yuanxin

机构地区：[1]西北大学城市与环境学院,陕西西安710127 [2]宝鸡文理学院地理与环境学院,陕西宝鸡721013 [3]宝鸡文理学院陕西省灾害监测与机理模拟重点实验室,陕西宝鸡721013

出　　处：《地理空间信息》2020年第12期5-9,I0005,共6页Geospatial Information

基　　金：国家自然基金资助项目(41771215);陕西省科技统筹计划项目(2016KTCL03-17)。

摘　　要：提出了一种基于哨兵二号2、3、12波段的增强阴影水体指数(ESWI),将其与归一化水体指数(MNDWI)相结合形成混合水体指数(MWI)模型,在同时具有平原及山地背景下的宝鸡市进行复杂地域水体提取实验。研究结果表明ESWI能够有效提取出水体并过滤掉阴影信息,MWI对于水体及其他地物的总体分类精度达到79%,Kappa系数达到0.695,提高了水体、非水体以及阴影之间的可分离性,剔除了水体以外的无关背景信息,有效地解决了水体与其他地物的混淆问题,适用于复杂地域水体的快速提取,可以用于大范围水体信息普查工作。In this paper,we proposed an enhanced shadow water index(ESWI)based on Band 2,Band 3 and Band 12 of Sentinel-2 and a mixed water index(MWI)model formed by the combination of ESWI and modified normalized difference water index(MNDWI)to realize effective extraction of water body in complex area.And then,we carried out a water body extraction experiment in complex area in Baoji City where there were both plains and mountains.Results show that ESWI can effectively extract water body and filter out shadow information.MWI has an overall classification accuracy of 79%for water body and other ground objects.The separability between water body,non-water body and shadow is improved with Kappa coefficient of 0.695.The irrelevant background information outside the water body is effectively eliminated,and the confusion between water body and other ground objects is solved.The proposed indexes are proved to be applicable to rapid extraction of water body in complex area and can be used for large-scale water body information survey.

关 键 词：复杂地域 水体提取 MNDWI MWI 阴影 

分 类 号：P237[天文地球—摄影测量与遥感]