| Information on the distribution and variation of atmospheric water vapor is essential for meteorological applications.Nowadays,the most commonly used technologies for measuring atmospheric water vapor have many problems remain to be solved(e.g.,low spatial and temporal resolution,high fabrication and operation cost,the accuracy of the results affected deeply by the weather and limitation of observation range).These problems all limit the development of the atmospheric water vapor remote sensing with high spatial and temporal resolution.With the development of Global Navigation Satellite Systems(GNSS),using GNSS measurements to remotely sense water vapor in the atmosphere has attracted significant attention due to their 24 hours availability,global coverage and low cost.GNSS tomography is a promising means of measuring and monitoring of water vapor in the troposphere.Its function is to build the three-dimension distribution of water vapor above the GNSS network in high resolutions.Nowadays,the commonly used tomographic model is not consistent with the fact that the distribution of GNSS signals(i.e.,the observation area)quickly changes over time.Moreover,the above problem will also result in too many unknown parameters and thus to degrade the performance of the tomographic solution.To address the issues which exist in the conventional tomographic approaches,key technologies for GNSS water vapor tomography,i.e.,determination of the tomographic region,discretization of the tomographic region and tomographic inversion,were studied.The main researches are as follows:1)To remove the unnecessary region in GNSS tomographic model,the principle for determination of the tomographic region was proposed by study the theory and method of GNSS tomographic model.Based on this principle,the redundant parameters existed in the unnecessary region have been removed effectively.Moreover,it can improve the accuracy and instability of the conventional tomographic model.2)The adaptive voxel-based GNSS tomography was proposed.The voxel was used for the discretization of the tomographic region and the projection plane algorithm is used for determination of the voxels,which are crossed by GNSS signals.Since the GNSS signals are ever-changing through the time,the GNSS tomographic regions combined by these voxels are also ever-changing at different tomographic epochs to adapt the distribution of the GNSS signals.Since water vapor exponentially decreases with the increase of height in the troposphere,thus in tomographic modeling,the nonuniform vertical structure was used.The results show that the new approach improved in comparison to the traditional approach.3)The adaptive node-based approach was proposed.The node was used for the discretization of the tomographic region and the Graham scan is used to discrete the tomographic region(three-dimensional structure)into tomographic boundaries(two-dimensional structure)at different heights.Meshing techniques are used to discrete the values of water vapor into the parameters at equally distributed nodes in each tomographic boundary.Due to the variations of water vapor,the vertical structure of the tomographic region was spaced unevenly.The results show that the new approach improved in comparison to the traditional approach.4)The access order scheme based on the algebraic reconstruction technique(ART)was proposed for the tomographic inversion.This is different from the commonly used observation equation system in which the order of the equations does not need to be considered.The access order scheme is based on the accuracy and the amount of information to achieve the reliable tomographic results.For the ordering of the observation equations,the access order scheme based on prime number decomposition was adopted such that the observation equations between two consecutive iterations are largely uncorrelated.The results show that the access order scheme can adaptively order the tomographic equations at each tomographic epoch and the tomographic results are improved. |
PDF Full Text Request