An Adaptive Threshold Selected Method from Remote Sensing Image based on Water Index

Abstract

Water is one of the most common and important objects on the earth, and its extraction is of great significance to many related researches in remote sensing domain. Water index method is most commonly used, and the accuracy of image interpretation is an unavoidable problem. However, traditional interpretation methods are subjective, and the efficiency of interpretation is relatively low. An adaptive threshold selected method based on modified normalized difference water index (MNDWI) is proposed here to extract water completely and accurately from remote sensing image. First, the study compares the effect of three commonly used water index methods (NDWI, MNDWI and MBWI) to extract water in Baiyangdian watershed, the result shows that MNDWI is better in separating water from background; then, a probability index method integrated water Z-score (IWZ) is used to segment MNDWI results. It shows that the optimized result with IWZ is more accurate than the traditional result with 0 as the dividing point, the former is 5% higher accuracy than the latter.

Combining remote sensing and social media data for flood mapping: a case study in Linhai, Zhejiang Province, China

“…In recent years, a large number of studies have focused on flood mapping 3 – 5 Depending on the time of data collection, flood inundation mapping is generally divided into two modes: real-time (RT) and near RT (NRT) mapping. RT mapping utilizes RT data from water level sensors, timely official flood hazard reports, or crowdsourced data, whereas NRT mapping uses data shortly after a flood event, usually with a lag of several days.…”