水汽层析代数重构算法  

Algebraic Reconstruction Algorithm of Vapor Tomography

作　　者：何林 柳林涛[2] 苏晓庆[2,3] 许超钤[4] 段鹏硕[2,3] 

机构地区：[1]贵州电力设计研究院,贵州贵阳550000 [2]中国科学院测量与地球物理研究所,湖北武汉430077 [3]中国科学院大学,北京100049 [4]武汉大学测绘学院,湖北武汉430079

出　　处：《测绘学报》2015年第1期32-38,共7页Acta Geodaetica et Cartographica Sinica

基　　金：国家自然科学基金(41074050;41021003)~~

摘　　要：讨论了代数重构算法在水汽层析应用中的各种问题,包括约束条件的构造、层析初值的选择、松弛因子的计算、终止条件的确定等,给出了计算最优松弛因子的黄金分割搜索法和确定终止条件的NCP规则,对比分析了Kaczmarz、Randkaczmarz、Symkaczmarz、SART、Landweber、Cimmino、CAV、DROP等8种常见的代数重构算法,并以香港SatRef的观测数据进行了试验。试验结果表明,以上8种代数重构技术都能够满足水汽层析的要求;迭代终止条件比松弛因子更为重要;采用文中计算最优松弛因子的黄金分割搜索法和NCP迭代终止条件,CAV算法结果最优,其次为Cimmino算法。While applying algebraic reconstruction algorithm in vapor tomography ,problems have to be solved with respect to constructing the constraint condition ,selecting the initial value ,calculating optimal relaxation factor and deciding the iteration termination condition .Golden section search method and NCP termination rule are given to solve the latter two problems ,respectively .Eight algebraic reconstruction algorithms ,including Kaczmarz ,Randkaczmarz ,Symkaczmarz ,SART ,Landweber ,Cimmino ,CAV and DROP algorithm ,are comparatively analyzed and tested by the data fromSatRef station in Hong Kong .The results show that all the eight algorithms can satisfy the requirements of vapor tomography and the iteration termination condition is more important than the relaxation condition .While the golden section method and NCP method are used ,the CAV algorithm performs best ,and then the Cimmino algorithm .

关 键 词：代数重构 水汽层析 GPS气象学 

分 类 号：P228[天文地球—大地测量学与测量工程]