Algebraic Reconstruction Algorithm of Vapor Tomography

Abstract

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 from SatRef 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.