GNSS Water Vapor Tomography Refined Models Considering Water Vapor Distributions And Software Development

In recent years,extreme weather events have occurred frequently,heavy rainfall and typhoons lead to natural disasters such as urban flooding,waterflood and mountain torrents.It’s urgent to develop high-precision atmospheric water vapor monitoring technique with high temporal and spatial resolution to provide scientific and effective basis for forecasting extreme weather such as typhoon and rainstorm.At present,the Global Navigation Satellite Systems(GNSS)water vapor tomography technique has become an important method to retrieve high spatiotemporal atmospheric water vapor distribution.,due to its high precision and all-weather availability.The discretization modeling of 3D tomographic region is the core step and research hotspots of GNSS water vapor tomography technique.The common method divides the 3D tomography area with a uniform discretization scheme,using the same horizontal division for all tomography height layers.However,this method does not follow the actual distribution of atmospheric water vapor in the vertical direction,nor does it agree well with the spatial distribution of GNSS signals.This seriously affects the stability of the tomography model and the accuracy of the results.This thesis focuses on the shortages of the discretization method in the current GNSS water vapor tomography technology,and systematically analyzes the vertical variation characteristics of atmospheric water vapor,and improved methods of GNSS water vapor tomography.The detailed work and research contents are outlined as follows:(1)Based on the key influencing factors of GNSS water vapor tomography technique,a systematic and comprehensive analysis was conducted on the impact of initial observations and constraints on the observation stability and inversion accuracy of GNSS water vapor tomography.The experiments based on the Hong Kong indicate that using radiosonde data to calculate water vapor densities and using them as initial observations can obtain reliable results;Compared with the high-accuracy prior information,the method using exponential function as the vertical constraint is more universal when the accuracy changes little;Horizontal constraints utilize the principle of image smoothing,which may sacrifice some accuracy,but can appropriately reverse the water vapor parameter information of blank voxels,compensate the shortages of the geometric observation structure of the water vapor tomography model.Therefore,it is necessary to add horizontal constraints in the tomography.(2)Taking the vertical distribution of water vapor as the breakthrough point,the inversion accuracy of GNSS water vapor tomography model under different discretization schemes is compared.This thesis studies the number of effective signals and the accuracy of tomography in different layer top height schemes.Experiments show that when the layer top height is set high,the number of effective signals drops sharply.However,because this scheme divides too many layers to the tropospheric top area with small absolute error,the accuracy is significantly improved compared with other lower layer top schemes.The experiments based on the same measured GNSS data show that,in different discretization schemes,the accuracy of the tomography results is similar in the tropospheric top zone where water vapor is scarce and changes slowly,but in the near ground layer where water vapor content is rich and changes violently,the accuracy of the low resolution discretization experiments has significantly improved.(3)Taking full account of the above comparative analysis and the vertical variation of water vapor,a GNSS non-uniform discretization water vapor tomography method considering the water vapor distribution is proposed.This method combines the spatial characteristics that water vapor is abundant and changes violently in the lower layer,and is sparsely distributed and changes slowly in the upper layer,and constructs a new non-uniform discretization scheme that gradually reduces the spatial resolution from the surface to the top troposphere.Experiments show that this method can effectively improve the penetration rate of 3D voxels in the tomographic region,and significantly improve the accuracy of the results.It shows that the non-uniform discretization GNSS water vapor tomography method can effectively improve the accuracy and quality of the atmospheric water vapor tomographic results while ensuring the resolution of the tomographic region.It is expected to provide a 3D atmospheric water vapor product with higher accuracy and higher resolution near the surface for rainfall prediction.(4)In view of the distribution of 3D atmospheric water vapor and the demand for fine monitoring,the GNSS adaptive discretization water vapor tomography software is further developed.This software can flexibly set different discretization methods of GNSS station network and tomography model,and users can also freely select various parameters such as different satellite system data,initial observation values,tomography epoch,constraints,etc.At the same time,this software provides users with multiple precision evaluation reference values,which can effectively analyze multiple schemes,assist users in designing the optimal GNSS water vapor tomography model based on actual needs,and achieve visualization of 2D water vapor profiles and 3D water vapor fields,making it convenient for users to monitor the spatial distribution of water vapor in a refined manner.The thesis has 39 figures,17 tables,and 89 references.