Study On Retrieval Method Of Soil Moisture And Salt Content In Farmland Using Multi-Source Remote Sensing

Soil moisture and salt content information is an important part of geo-information system,and is also one of the key parameters in the field of earth science to study the energy flow of each sphere.Understanding soil moisture and salt content is of great significance to climate,geological evolution and ecosystem balance.At the same time,real-time acquisition of soil moisture and salt content in large areas of cultivated land can provide guidance for ecological environmental governance and the implementation of precision agriculture.With the emergence of multi-source and high-resolution remote sensing,the reliability of rapid and large-scale soil moisture and salt monitoring has been greatly improved,but there are still many difficulties and uncertain characterizations of water and salt retrieval at the field scale.At present,the problems of soil moisture retrieval using remote sensing at the field scale are mainly shown as follows:1)The applicability of existing soil moisture retrieval models is poor.Most of the models are based on microwave radiation transfer theory and are expressed as forward models of backscattering coefficient.Using this kind of model to simulate soil moisture content is easy to lead to the accumulation of various errors into soil moisture,resulting in the decline of accuracy.2)The surface roughness parameters in the models have a great impact on the accuracy of the models.The surface roughness of the field plots is highly variable,subject to human interference,and more complex than the process of determining moisture content in the field.3)The existing models are mostly established for non-salinized areas,and the emergence of salt will lead to a certain degree of change in the soil dielectric constant.At present,there is no effective moisture content retrieval model for the soil of saline land.4)The application of multi-source remote sensing data is insufficient.Optical remote sensing is sensitive to surface moisture,but it is easily disturbed by surface factors such as evaporation and vegetation,which will lead to the failure of the retrieval models.Aiming at the above problems,this paper takes the bare land,the saline bare land,and the crop cover land as a typical research area and analyzes the coupling relationship between soil moisture and salt content.We establishe a retrieval model of different type of cultivated land.The main research results are as follows:(1)On the basis of the analysis of the backscattering and scattering mechanism in the saline soil area,differences between the backscattering of salt accumulation and non-salt accumulation surface are found.The extraction methods of accumulated salt and nonaccumulated salt cultivated land are put forward.Based on the statistical analysis of various salt indexes,the salt index and threshold of salt extraction are defined to salt accumulation area.We establish an empirical model for retrieving salt content in non-accumulated salt area.(2)A method of soil roughness division using scattering entropy is proposed,so the cultivated land surface can be divided into Bragg surface and rough surface.By analyzing the soil water and salt of the Bragg surface,the co-polarization ratio retrieval model for nonsalt accumulation surface,the cross-polarization ratio inversion model and T model for the accumulated salt area are established.These models eliminate the influence of surface roughness and can achieve an retrieval accuracy of less than 0.05cm3/cm3.(3)The surface roughness parameters,the third eigenvalue(λ3)of rough surface is proposed,and the effectiveness of λ3 is verified using the Oh model.The availability of first eigenvalue(λ1)and second eigenvalue(λ2)are analyzed.On this basis,a soil moisture retrieval model based on vertical polarization backscattering coefficient(σVH0)is proposed which has a good applicability.The retrieval accuracy of the model can reach 0.042 cm3/cm3.(4)An improved Dubois model for non-saline bare land is proposed,which is based on crosspolarized(VH)and vertically polarized(VV)SAR images.On the basis of the sensitive analysis of the backscatter coefficients and soil moisture,the influence of factors such as angle of incidence,rainfall and soil texture on the backscatter coefficient is discussed.The model improves the Dubois model through terrain radiation correction and introducting empirical formulas of surface roughness,which eliminates the influence of the angle of incidence and solves the problem that the surface roughness needs to be measured in the field.The experimental results show that the modified Dubois model can effectively retrieve soil moisture under the conditions of multiple soil textures and angle of incidence,and has the higher accuracy than the original model.(5)A modified water cloud model under crop cover surface is proposed.Firstly,multiple vegetation indices are analized,which are used as vegetation moisture in the water cloud model.Then,the variation of bidirectional attenuation coefficients under different Normalized Difference Vegetation Index(NDVI)conditons is analized.We discusse the influence of surface roughness and soil texture on the backscatter coefficients,the surfaceroughness is corrected using cross-polarization ratio(σVH0/σVV0)and Transformed Soil Adiusted Vegetation Index(TSAVI)in the model.The retrieval accuracy of the modified model is significantly improved compared with the original water cloud model.