摘要
Abstract
As a set of automated process has not yet been formed for constructing reservoir capacity curves using remote sensing technology,it is proposed to determine the optimal threshold with the inflection point method of the cumulative frequency curve based on the water body index method.The edge of the water body is optimized by the regional seed growth algorithm,so as to construct a complete set of automated process to quickly construct the reservoir capacity curves.With Sentinel-2 remote sensing images as the data source and Chegu Reservoir in Wuan City,Handan City,Hebei Province as the research object,this paper uses five land and water image metrics,namely,NDWI,MNDWI,RWI,MBWI,and AWEInsh to extract the area of the water body,construct the reservoir capacity curve,and compare it with the results of the measurements in 2017.The results are as follows.The extraction accuracy of NDWI and RWI water body index is high.The average error of Chegu water body area extraction is 2.5%.The relative error is-7.3%～5.8%and-6.2%～4.7%,and the R² is 0.993 36 and 0.990 49,respectively.The range of the relative error of the reservoir capacity is-9.4%to-1.8%and-9.3%to-1.7%,respectively.The study shows that the cumulative frequency curve inflection point method is highly applicable and stable in determining the optimal threshold of water body index,with a simple and feasible principle.The regional seed growth algorithm can effectively eliminate the discontinuity at the edge of the water body,reduce systematic error,and further improve the accuracy.The cumulative frequency curve inflection point method provides a new idea for remote sensing water body extraction,serving as a reference for the batch construction of national small and medium-sized reservoir capacity curve extraction.关键词
Sentinel-2/水体指数/累积频率曲线拐点法/生长算法/库容曲线/车谷水库Key words
sentinel-2/water Index/inflection point method of the cumulative frequency curve/growth algorithm/storage capacity curve/Chegu Reservoir分类
建筑与水利
