Font Size: a A A

Microwave Remote Sensing Estimation Of Soil Moisture In Wheat Fields In Coastal Saline Soil

Posted on:2020-12-10Degree:MasterType:Thesis
Country:ChinaCandidate:L H DingFull Text:PDF
GTID:2480306005469764Subject:Agricultural information technology
Abstract/Summary:PDF Full Text Request
Water is an important source of life on earth.Soil moisture is one of the basic conditions for crop growth,an important indicator for crop growth monitoring,crop yield estimation and farmland drought monitoring,and a key factor for the water and energy cycle of terrestrial ecosystems.The traditional method of measuring soil moisture content is time-consuming and laborious,which is not conducive to the monitoring of soil moisture in a large area.Microwave remote sensing is able to work all day and all-weather,and can penetrate the clouds,and has a certain penetration ability to the vegetation layer,providing a possibility for real-time,dynamic and large area monitoring of soil moisture.Based on previous studies at home and abroad,the feasibility of using microwave remote sensing combined with optical remote sensing to estimate soil moisture in saline soil area was analyzed,and the development trend of using microwave remote sensing combined with optical remote sensing to estimate soil moisture in coastal saline soil area was forecasted.Based on Sentinel-1 microwave remote sensing data,Sentinel-2 optical remote sensing image and ground water content data,this study estimated the soil water content in the wheat field of coastal saline soil using the western wudi county of shandong province as the research area,and established a model for estimating the soil water content in the wheat field of coastal saline soil.First,sentinel-1 microwave remote sensing data was preprocessed to extract the radar backscattering coefficient.Meanwhile,Sentinel-2 optical remote sensing images were processed for atmospheric correction.The normalized vegetation index and salinity index were constructed by using the images processed by atmospheric correction.Then,based on the normalized vegetation index,the vegetation water content was estimated and substituted into the water cloud model to reduce the influence of vegetation scattering,so as to obtain the direct backward scattering coefficient of soil,and the salinity index with the highest correlation with the soil water content in the study area was selected.Finally,the estimation models of soil moisture were constructed and verified based on the radar backscattering coefficients and the direct backscattering coefficients of soil and the optical auxiliary parameters of different polarizations(combinations).Several models were compared to select the best soil moisture content estimation model.The research results are as follows:(1)the effect of vegetation layer on soil backscattering was reduced by water cloud modelIn this study,Sentinel-1 data of microwave remote sensing and Sentinel-2 images of optical remote sensing were used to estimate the soil moisture content in the study area.A water cloud model was used to estimate the influence of vegetation layer on backscattering.Sentinel-2 optical remote sensing data was used to construct vegetation index to estimate vegetation moisture content(mveg),which was used as a parameter of the water cloud model to calculate vegetation double-layer attenuation coefficient and reduce the influence of vegetation layer backscattering,so as to obtain the direct backscattering coefficient of soil in the study area.The results show that the direct backscattering coefficient of soil obtained by the two polarization modes through the water cloud model has a different degree of decline compared with the original radar backscattering coefficient.The VV polarization backscattering coefficient decreases from about-15d B?-6d B to about-17d B?-7d B,with a decrease range of about 1.5d B.The backward scattering coefficient of VH polarization mode decreased from about-24d B?-16d B to about-26d B?-18d B,with a decrease range of about2d B.It is known that the influence of vegetation layer on VH polarization backscattering coefficient is greater than that of VV polarization backscattering coefficient.Can also be learned that after the soil of the radar to the influence of the scattering coefficient greater than the low vegetation to after the radar scattering coefficient,the influence of electromagnetic waves of Sentinel-1 satellite launch in low vegetation cover penetration ability is stronger,therefore,can make use of Sentinel-2 data to construct the optical vegetation index of vegetation water content and its substitution waterclouds model to cut after the vegetation to the scattering effect,and can obtain a certain effect.(2)backscattering coefficients of different polarization combinations are constructedBy considering the backward scattering field of the earth's surface characteristic and polarization of Sentinel-1 image,the polarization mode of microwave remote sensing image to different forms of combination to abate the noise to the influence of the scattering coefficient,after building the VV,VH,VV/VH,(VV+VH)/(VV-VH)and other four kinds of polarization and the soil moisture estimation model(combination)mode to join the normalized difference vegetation index(NDVI),salt index(NDSI,S5,SI)as an auxiliary parameter of inshore saline soil area to evaluate the soil moisture content wheat fields.(3)the estimation model of soil water content was established and verifiedThe soil moisture estimation model based on microwave remote sensing,the soil moisture estimation model based on water cloud model and the soil moisture estimation model based on salinity index and water cloud model were established respectively.The unitary linear regression,support vector machine regression and partial least squares regression are selected to establish the model.Among them,the(VV+VH)/(VV-VH)+NDSI+S5+SI combined SVM model based on salinity index and water cloud model has the best soil moisture estimation results,and the R2of modeling and validation determination coefficient is the highest among all models(R12=0.7713,R22=0.5121),and the RMSE of modeling and validation is the lowest among all models(RMSE1=0.0407,RMSE2=0.0599).Shows waterclouds validity to reduce the influence of vegetation scattering model,add salt index to remove soil salt has obvious effect on soil moisture estimation effect(the model precision is increased from 0.5 to 0.7),and(VV+VH)/(VV-VH)polarization method and support vector machine(SVM)regression method superiority in soil moisture estimation.It also shows the validity of sentinel-1 and sentinel-2 images in the estimation of soil water content.To sum up,Sentinel-1 microwave remote sensing data combined with Sentinel-2 optical remote sensing images provide a new method for the wide range and dynamic monitoring of soil moisture content in coastal saline soil areas,providing an important theoretical basis and technical support for the treatment of salinization and improvement of the ecological environment of saline land areas.
Keywords/Search Tags:Sentinel-1, Sentinel-2, soil moisture, polarization combination, water-cloud model, salt index
PDF Full Text Request
Related items