Research On Rice Identification In Cloudy Regions Based On Time Series SAR Imagery And InSAR Technology

As the main food of nearly half of the world's population,rice can be used not only as edible,but also as raw materials for brewing,sugar and other industries.With the development of global economy and the rapid growth of population,a large number of cultivated land is occupied.The deterioration of rice habitat environment caused by human activities leads to the reduction of planting area and total yield of rice and other food crops.The contradiction between human and land is prominent.The lack of food supply will inevitably lead to hunger and poverty,which will hinder the development of social economy.At the same time,the management and improvement of rice habitat environment,such as water and soil management,pest control,energy and waste management,need to monitor rice growth.And a large number of studies show that the global rice planting area is rapidly decreasing.The use of remote sensing technology to monitor agriculture is currently the mainstream method.When monitoring rice planting area,the use of optical remote sensing image is limited due to more clouds,while synthetic aperture radar(SAR)is the main method Radar(SAR)is widely used in agricultural monitoring because of its high resolution,multi-mode,multi polarization direction,data acquisition process independent of the natural light source,which has the characteristics of penetrating cloud and fog,all-weather,all-time land observation and sensitive to the shape of ground objects.It is one of the important means to monitor rice in the area of multi cloud and fog.This study analyzes the polarization backscattering coefficients of main features on the time-series SAR images during the rice growing period,and combined with InSAR technology,the recognition results of rice areas with three schemes of single time phase SAR image,multi time phase SAR image and combination of multi time phase SAR image and InSAR technology are obtained through experiments,which are verified and analyzed from two aspects of sample point distribution and extraction area,The rice recognition accuracy of the case is compared with that of sentinel-2 optical image,so as to provide suggestions for using SAR remote sensing technology to identify rice planting areas in cloudy areas.The main conclusions of this study are as follows:(1)Based on InSAR technology,the VV and VH polarization coherence coefficients of typical ground features in rice growing period are obtained.Through analysis,it is found that the VV polarization coherence coefficients of artificial buildings are greater than that of VH polarization coherence coefficients,while the VV and VH polarization coherence coefficients of rice,vegetation and water bodies are similar.However,it can be found that VV and VH polarization coherence coefficients can well distinguish artificial buildings and other features.In August and September,the VV polarization coherence coefficients of rice can well distinguish rice from water and vegetation.In August and October,the VH polarization coherence coefficients of rice can also well distinguish rice from water and vegetation.Therefore,this study believes that the polarization dryness introduced by InSAR technology can well distinguish rice from water and vegetation Correlation graph(coherence coefficient)can be used to identify rice.(2)It is found that,similar to previous studies,the rice recognition accuracy of time series sentinel-1 SAR image is lower than that of time series sentinel-1 SAR image combined with InSAR technology,with the user accuracy of 78% and 81% respectively.The rice recognition accuracy of time series sentinel-1 SAR image combined with correlation coefficient is higher than that of time series sentinel-1 SAR image increased by 3%.However,the accuracy of radar image classification is still lower than that of optical image.(3)The Sentinel-2 optical remote sensing image has higher spatial resolution(10m),and the image acquisition time is about 20 days after rice planting.On sentinel-2 image,the rice area has obvious characteristics difference with other features,so this study directly uses object-oriented classification method to extract rice area,and uses sampling points and statistical area to test respectively It is proved that the accuracy of point data verification is 93.50% and the accuracy of area verification is 91.76%.According to the previous research,the remote sensing classification results with the accuracy of 90% are very good.Therefore,this study believes that the data extracted from sentinel-2 can provide a reference for the analysis of rice area recognition results based on time series sentinel-1.(4)Based on the results of sentinel-2 optical remote sensing image rice classification,the comparative analysis is based on sentinel-1 The three rice classification results of SAR image show that time series SAR image can obviously distinguish rice and vegetation information and improve rice recognition accuracy compared with single time phase SAR image,while the introduction of coherence coefficient not only improves the classification accuracy of artificial buildings,but also clearly distinguishes vegetation and rice,thus improving rice recognition accuracy.(5)Considering the advantages and disadvantages of optical and SAR images,in most areas,including cloudy areas,when monitoring crops such as rice,if there is available optical data,optical data is preferred.If there is a lack of optical data,SAR data of synthetic aperture radar can be used for monitoring.At the same time,the image width of sentinel-1 series reaches 25 This study also suggests sentinel-1 SAR data to be used in large-scale rice monitoring or mapping in plains or areas with small topographic relief.In a word,this study takes Guanghan City,the core city of Chengdu Plain,as an example to study the time series sentinel-1 SAR image and InSAR technology are combined to extract rice area,and good results are obtained,which proves that the coherence coefficient has an auxiliary role in rice area identification;research methods and results have reference significance for rice planting area extraction in cloudy area,at the same time,this paper only identifies rice area,while rice related habitat environment management and improvement are achieved Good,such as water and soil management,pest control,and energy and waste management,are also worthy of further study.