SAR Flood Monitoring And Analysis For Dongting Lake Area

With the drastic change of global climate,the frequency of heavy rainfall events is increasing,and the destructiveness and losses caused by floods are increasing year by year.In order to ensure the safety of lives,property and public facilities,fast and accurate flood detection methods are of great significance in disaster early warning,flood relief and post disaster reconstruction.Traditional flood detection methods mainly rely on artificial investigation and station monitoring,which is difficult to respond to flood changes in time.With the development of remote sensing earth observation technology,the launch of a large number of remote sensing satellites provides important data support for flood detection technology.However,the traditional optical remote sensing imaging technology is easily affected by extreme weather such as clouds and rainfall in flood disasters,so it is unable to obtain high-quality images.Synthetic aperture radar(SAR)is widely used in flood detection methods because of its active imaging characteristics,and the image obtained is hardly affected by cloud and fog.Aiming at the flood event of East Dongting Lake in 2020,this paper designs two intelligent flood detection methods based on change detection and time series analysis,and verifies the effectiveness and reliability of the proposed method through experiments.The details are as follows:(1)In order to achieve fast and effective flood extraction,this paper proposes a flood extraction algorithm of dual-polarized Sentinel-1 SAR image based on change detection.This method makes full use of the information of two different polarization modes of Sentinel-1,fuses the difference map of flood change under different polarization modes through the change index,and supplements the edge information of the fused difference map to realize the accurate extraction of flood area.(2)In order to overcome the problem that the change detection algorithm relies too much on artificial selection of images before flood events,a new flood extraction framework based on spatio-temporal feature fusion of dual-polarized Sentinel-1 SAR images is designed in this paper.Different from the existing flood detection methods,flood extraction is often carried out from the perspective of a single spatial domain or time domain.This method fully exploits the application potential of dual-polarized time series data and realizes the feature construction and fusion extraction in line with the temporal and spatial characteristics of flood.And this method can generate flood mapping results for the whole flood event and different flood moments in the flood event at the same time,so as to meet the needs of more comprehensive flood analysis.(3)Design and application of flood mapping software.This paper is based on Python3.7and Py Qt Visualization Toolkit,completed the software design of flood mapping,and tested the constructed flood event data set.In addition,the software also adds the function of analyzing the features affected by flood,which can complete the spatial and statistical analysis of the disaster situation of the features in the observation area according to the flood extraction results.