Study On Soil Moisture Inversion Methods Based On GPS Reflection Signal

Soil moisture is one of the key parameters to study the environmental and ecological treatment of mining area,and it is also the link between surface water and groundwater.It is an important part of water cycle.The rapid monitoring of soil moisture is of great significance in the fields of atmosphere,hydrology and ecology.Drying method is inefficient,time-consuming and laborious,and cannot be monitored in large area.The monitoring accuracy of visible light and thermal infrared remote sensing is easily affected by factors,such as atmosphere and upper vegetation.Space-borne microwave remote sensing data,such as SMOS and SMAP,are sensitive to changes in soil moisture and have certain penetrability to vegetation,but because of their low spatial resolution,they cannot meet the needs of regional small-scale monitoring.The L band electromagnetic wave signal transmitted by GNSS satellite has the advantages of all-weather,strong penetration and cheap and easy to obtain,and the reflected signal is very sensitive to the change of soil moisture.Therefore,it has become a new research hotspot in the field of monitoring the change of soil moisture.However,the traditional single-antenna GNSS-R soil moisture inversion model has the problems of poor stability and low inversion accuracy,and it is easy to ignore the diferences and complementarities between different satellite dual-band reflection signals.Aiming at these problems,a single-antenna model GNSS-R soil moisture foundation experiment was carried out in an open space on the east side of the north gate of Anhui University of Science and Technology.The single-satellite single-band soil moisture inversion method and dual-band soil moisture inversion method were studied.The main results are as follows:(1)In order to collect experimental data,the GNSS-R soil moisture ground-based experiment with single antenna observation mode was carried out.?The suitability of soil moisture monitoring in L1 band of PRN23 satellite is analyzed.The results show that the detection range of multipath reflection signal in PRN23 satellite L1 band is inversely proportional to its altitude angle.When the height angle is 5°,the detection range is 175.46m~2,the detection depth is about 1.92cm to 6.59cm.?It is necessary to analyze the correlation between the interference characteristic parameters of PRN16?PRN18?PRN22?PRN23 satellite and the measured values of soil moisture to provide the basis for the selection of model independent variables.The results show that phase of PRN16 L1 band fails significance test,and the phase of PRN23 satellite L2 band is the best correlation with the measured value of soil moisture(R=-0.8180),The correlation between amplitude and frequency and soil humidity of L1 and L2 of satellites is better than phase.(2)BP neural network algorithm and support vector machine(SVM)algorithm are used to improve the accuracy and reliability of the traditional single satellite single frequency soil moisture inversion model.The results show that:?In terms of the accuracy and reliability of the inversion models,the BP-traditional model and the SVM-traditional model are superior to the traditional model,and the accuracy is improved to different degrees.For example,R of the BP-the traditional model increased by 63.42%,RMSE of the BP-the traditional model reduced by 36.83%.R of the SVM-traditional model increased by 60.26%,RMSE of the SVM-traditional model reduced by 28.00%.?The inversion accuracy of each model in L1 band of PRN22 satellite is basically higher than that of PRN16 and PRN18 satellite.The amplitude-model of PRN18 satellite is better than that of PRN22 satellite.(3)In order to make full use of the difference and complementarity between L1 and L2 band reflection signals of different satellites,we use entropy fusion method,mean fusion method and adaptive fusion method,the amplitude value of each satellite L1 and L2 band is weighted fusion,and the dual-band soil moisture inversion model is established and verified.The difference between single-band soil moisture inversion methods and dual-band soil moisture inversion methods are compared from qualitative and quantitative angles.Studies have shown that:?Judging from the modeling stability,the order from high to low is adaptive fusion method,entropy fusion method(PRN22 satellite),mean fusion method,entropy fusion method(PRN23?PRN16?PRN18).?Soil moisture inversion model of adaptive fusion method is better than the mean fusion method and entropy fusion method,and R2 is 0.8727 and RMSE is 0.2551%.?From the qualitative point of view,the inversion accuracy of the traditional model is limited,and it is easy to ignore GPS differences and complementarities between reflected signals in different satellite L1 and L2 bands.The improved model based on BP neural network and support vector machine algorithm use the correlation between interference characteristic parameters.The dual-band soil moisture inversion method combines the difference and complementarity of amplitude between satellites and bands.?From the quantitative point of view,the single satellite single-band soil moisture inversion methods based on BP neural network and support vector machine and the dual-band soil moisture inversion methods can effectively improve the inversion accuracy of the traditional model.The inversion accuracy of adaptive fusion method is higher.Figure[31]Table[12]Reference[76]...