Research On Soil Moisture Retrieval Over The Agricultural Field Using Microwave Remote Sensing Data

Soil moisture is a key state variable that influences both global water and energybudgets.It plays an important role in the surface energy exchange and the dynamic change ofclimate. Therefore, soil moisture monitoring in large scale is great significance for weatherforecasting, flood and drought monitoring, crop yield estimation and so on.It is difficult to achieve in large-scale to measure soil moisture by traditional point-basedmethod. And this method is time-consuming and labor-intensive. Due to the limits of cloudcover, aerosol, solar radiation conditions and vegetation cover factors, the accuracy of soilmoisture measurement by optical remote sensing is very limited. Microwave remote sensinghas become one of the most effective means to monitor soil moisture rely on the advantagesof all-weather, day and night, could penetrate surface by now.The study includes two contents of based on radar data to retrieve soil moisture of thecorn field and based on radiometer data to study the crop radiation characteristics. Thefollowing will sum them up separatelyIn the field of active microwave remote sensing, there is a certain correlation between thebackscattering coefficient obtained from radar and the surface dielectric constant,Andfurthermore, surface dielectric constant apparently responses to the changes of soilmoisture.these provide a theoretical base for studying soil moisture inversion by using activemicrowave remote sensing. However, due to the complexity of radar and surface interactions,Surface roughness, radar instrument parameters, vegetation cover and other factors affect theradar signal as well. The studies showed that, similar with the soil moisture, radar also hasbeen sensitive about the information of vegetation layer. For this reason, it is necessary toremove the vegetation backscattering effect when the inversion of soil moisture in thevegetated areas. Using the improved first order microwave coherent scattering model basedon GIMICS and AIEM theory model to simulate the backscattering characteristics of cornfield, explored the relation between the bare soil backscattering and the total of vegetationbackscattering of numerical calculation (ratio and residual) and the vegetation parameters. Atthe same time, it has analysis the trend of backscattering characteristics with the vegetationparameters. Ultimately, choice the measurement data of two days which are June19andAugust21and successfully establish a semi-empirical model of bare backscattering andvegetation backscatter ratio and vegetation parameters.When the input parameters and output parameters are unknown, the neural networkalgorithm is the preferred for the nonlinear relationship. The study base on the simulation ofAIEM model, Put the correlation length, rms height and the backscattering coefficient of HHpolarization as the neural network input vector, soil moisture as the output vector to create theneural networks to retrieve soil moisture. Compared with the measured data of the soilmoisture, the retrieval accuracy is that R-square is0.6050, and the root mean square is0.00298.In addition, the study also processes and analyses the measured wheat data of Baoding,Hebei province in the year of2009, and uses the AIEM theoretical model and the methods of regression analysis and neural network to retrieve the soil moisture of wheat prophase. Theresults showed that the inversion result of the neural network method was superior to theinversion result of regression analysis method.