一种综合多特征和超像素的SAR湿地水淹频率计算方法

A Comprehensive Multi-feature and Superpixel SAR Wetland Flooding Frequency Calculation Method

湿地水域变化影响湿地生态功能发育,研究湿地水淹频率有助于湿地监测管理。为研究合成孔径雷达(SyntheticAperture Radar,SAR)多极化特征在湿地水淹频率中的应用,本文基于Sentinel-1雷达数据,提出一种综合多极化特征和超像素优化的SAR湿地水淹频率计算方法。该方法结合优化的雷达水体指数,采用统计投票,融合多极化特征和超像素空间上下文信息,能准确地识别水体,进而计算全年的湿地水淹频率。本文通过Google Earth Engine(GEE)平台,选取洞庭湖2020年的Sentinel-1 SAR数据展开试验,本文方法水体的提取精度为93.96%,计算的水淹频率分布与洞庭湖地势特点相符。

The changes in wetland water bodies affect the development of wetland ecological functions,and studying the frequency of wetland flooding is helpful for wetland monitoring and management.To investigate the application of multi polarization features of Synthetic Aperture Radar(SAR)in wetland flooding frequency.This article proposes a SAR wetland flooding fre‐quency calculation method based on Sentinel-1 radar data,which integrates multi polarization features and superpixel optimization.This method combines optimized radar water index,uses statistical voting,integrates multi polarization features and superpixel spatial contextual information,and can accurately identify water bodies,thereby calculating the annual wetland flooding frequency.This article uses the Google Earth Engine(GEE)platform to conduct experi‐ments on Sentinel-1 SAR data from Dongting Lake in 2020.The accuracy of the water body proposed by this method is 93.96%,and the calculated frequency distribution of flooding is consistent with the terrain characteristics of Dongtinghu lake.