Soil Moisture Remote Sensing With Global Navigation Satellite System Reflectometry

Although soil holds only a small percentage of the total global water budget,soil moisture plays an important role in the global energy and water cycle.It affects the transportation of energy and vapor by changing the surface albedo,and thus affects the climate change.Currently,several remote sensing and in-situ measurement techniques have been used for soil moisture monitoring.However,retrieving soil moisture with the required accuracy and the appropriate spatial and temporal resolutions still remains a major challenge.The use of Global Navigation Satellite Systems(GNSS)reflected signals as sources of opportunity for soil moisture measuring is assessed in this Ph D Thesis.The observation system,named GNSS-Reflectometry(GNSS-R),has gained increasing interest during the last two decades due to its unique characteristics.At present,the GNSS-R remote sensing research mainly focuses on LHCP reflected signals,the response of RHCP reflected signals to soil moisture variation were seldom mentioned and reported.Reflected signals on both circular polarizations were collected in the ground and airborne experiment,the study is mainly carried out from the following aspects:This study designed the SOMOSTA(Soil Moisture Monitoring Station)experiment on the intercomparison of soil moisture monitoring from GNSS-R signals and passive L-band microwave radiometer observations at the Valencia Anchor Station.The GNSS-R instrument has an up-looking antenna for receiving direct signals from satellites,and a dual-pol downlooking antenna for receiving LHCP(left-hand circular polarization)and RHCP(right-hand circular polarization)reflected signals from the soil surface.In addition,calibration was performed to reduce the impact from the differing channels.The power ratio among channels was concentrated around 1,with an average value of 1.03 and a mean absolute deviation of0.091 over the whole observation period.By comparing ground-based GNSS-R and ELBARA-II radiometer data,the correlation between LHCP/RHCP reflectivity by GNSS-R and HH/VV reflectivity by L-band radiometer was analyzed.A high correlation was found between the LHCP reflectivity measured by GNSSR and the horizontal/vertical reflectivity from the radiometer(with correlation coefficients ranging from 0.83 to 0.91).An airborne experiment was carried out at Shangjie Airport in Zhengzhou,Henan Province.Reflected signals on both circularly polarizations were received with the dual-pol nadir antenna.The received reflected signals were calibrated with water surface to eliminate the effects of atmospheric decay.The LHCP signals have the same variation trend as in field campaign,RHCP signals is insensitive to soil moisture variation.However,the ratio between both polarizations has higher sensitivity than LHCP signals at high elevations.For spacebone GNSS-R soil moisture observation,the reflected signals will be affected by atmospheric attenuation,in addition,the complexity of the underlying surface will also have a certain impact on the reflected signals.The reflected signals from TDS-1 and CYGNSS satellite-borne are analyzed,and the accuracy of the retrieval was improved by the antenna gain limitation and the quality control.The fading noise in the reflected signals is the main source of the retrieval error.Fading noise can be reduced by performing a large number of noncoherent integration.In this paper,the spatial distribution of GNSS reflected signals is analyzed.For L-band,the dependence of LHCP reflectivity with elevation is more obvious,while the azimuth effect could be considered negligible.On this basis,an empirical regression model was proposed,the LHCP reflectivity exhibits the same trend as the physical model in the range of 20-70° elevation angle,decreases with elevation angle,and increases with soil moisture.A Neural Net Fitting was proposed for GNSS-R soil moisture inversion.Daily averaged GNSS LHCP reflectivities at 10 different incidence angles was used as input,and in situ soil moisture measurements as target.The results clearly show that the algorithm is effective.Going still further,adding Rrr and MODIS EVI as input,significantly improved the accuracy of the results.