Soil Dielectric Property Analysis And Information Extraction Of Salt And Water In Sparse Vegetation Zone

Xinjiang Uygur Autonomous Region is located in the northwest of China.The spatial distribution of water resources is very unequal,resulting in more desert,less oasis.And there are most of regions of Xinjiang have less rainfall,dry climate,so the ecological environment is extremely fragile.At the same time,Xinjiang is located in mid-latitudes,rich in light resources,high crop yield and excellent quality.As a backup reserves of grain base in our country is responsible for food security,social security,national security and other major strategies.In recent years,due to the unreasonable regional farmland abuses such as development,groundwater resources,regional ecological environment worsening,especially in the oasis,desert crisscross area,large areas of land degradation and ecological diversity.Land degradation caused by many factors,but the lack of soil moisture,high salt content are oasis desert crisscross area-an important factor of land degradation.At present,the land salinization stress in Xinjiang area of about a third of the total area of arable land.Soil salinization phenomenon has been a serious threat to the coordinated development of regional ecological system and inhibited the normal growth and production of crops.In this paper,the typical oasis of Weigan-Kuqa oasis in Xinjiang is selected as the study area.The oasis stability is influenced by the water resources.Under the stimulation of the western development and other related policies,the social and economic level of the county and county in the Oasis has been increasing year by year and the agricultural land area has been expanded rapidly.The huge water demand has increased the spatial distribution of water resources.But surrounding the oasis ecosystem worsening,sparse vegetation coverage area of land degradation phenomenon,buffer zone between oasis and desert area has been gradually shrinking.The development of spatial sensing means realizes the large-area and multitemporal soil information extraction technology.The accuracy of soil information monitoring is increasing in the background of rapid development of relevant technical means.Especially the high-resolution remote sensing data,which provides a guarantee for the accuracy of soil information monitoring.The multi-polarized RadarSat-2 and GF-1 images were selected as the data sources.1)based on the real and imaginary part of dielectric constant of measurement,I used a variety of mathematical transformation forms to establish dielectric constant stepwise regression equations of soil water and salt;2)Using the AIEM model under a certain incidentangleto simulate the surface roughness,soil moisture and radar backscatter and established the bare surface soil moisture experience model;3)A variety of spectral indices of GF-1 are used.And using the BP neural network model for radar data and multi-spectral data to establish the model of soil salt and validation and accuracy of evaluation.The main conclusions are as follows:1.The correlation of soil moisture content and soil dielectric constant real component is 0.94.And the water content has a great influence on the real part.The imaginary part is controlled by water content and salt content.In the case of large salt,the greater the water content,the larger the imaginary part of the permittivity.It has a weak correlation between the imaginary part and the soil salinity,only 0.45,but highly correlated with water content,correlation can reach 0.90.In addition,the use of dielectric constant of 5 kinds of transformation form,soil moisture real component model is set up by high precision,the reciprocal model is the optimal model of soil moisture.The coefficient of determination R2 is 0.82 and the verification accuracy is0.91.At the same time,the soil salt and salt index of square(moisture + salt)were used as the optimal model by using the imaginary model.The coefficient of determination R2 was 0.84 and the accuracy of R2 was 0.91.Moisture and real part and imaginary part showed a good linear relationship,while the imaginary part and salt and imaginary part of a nonlinear relationship.2.The surface roughness and soil moisture were simulated by AIEM model.The regional roughness was consistent with the actual situation.The oasis-desert ecotone and the roughness near the river were larger.The soil moisture was higher and the coefficient R2 was 0.86,The soil water content is lower in the west and east,and the soil moisture in the oasis and the river is higher.3.The soil salt linear empirical model is difficult to accurately assess the soil salt content,so combined with multi-spectral auxiliary data on the backscatter information segmentation.Using four polarization of RadarSat-2 scattering coefficient and the surface roughness and soil moisture of radar data to extract spectral index of GF-1data sets.And build the BP neural network model.After many experiments,the BP neural network model is better to extract the regional soil salt information,model simulation accuracy reached 78.95%.From a certain extent,verify the optical remote sensing and active microwave remote sensing information extraction efficiency of saline area.In this paper,the empirical model of soil moisture inversion in bare soil was established by using the active microwave remote sensing feature.At the same time,the soil salinity model was established by using BP algorithm,and the simulation accuracy.The study can provide some reference for soil information extraction in arid area.