Quasi-Monte Carlo, Monte Carlo, and regularized gradientoptimization methods for source characterization ofatmospheric releases

Multi-Version Program Analysis

计算机科学;物理学

B Addepalli ; C Sikorski ; E R Pardyjak ; M Zhdanov

Others  :  http://drops.dagstuhl.de/opus/volltexte/2009/2299/pdf/09391.SikorskiKrzysztof.Paper.2299.pdf
PID  :  6410

学科分类：计算机科学（综合）

来源: CEUR

【 摘 要 】

An inversion technique comprising stochastic search and regularized gradient optimization was developed to solve the atmospheric source characterization problem. Theinverse problem comprises retrieving the spatial coordinates, source strength, and the windspeed and wind direction at the source, given certain receptor locations and concentrationvalues at these receptor locations. The Gaussian plume model was adopted as the forwardmodel and derivative–based optimization was preferred to take advantage of its simpleanalytical nature. A new misfit functional that improves the inversion accuracy of atmosphericinverse-source problems was developed and used in the solution procedure. Stochastic searchwas performed over the model parameter space to identify a good initial iterate for the gradientscheme. Several Quasi-Monte Carlo point-sets were considered in the stochastic search stageand their best performance is shown to be 5 to 100 times better than the Mersenne-Twisterpseudorandom generator. Newton's method with the Tikhonov stabilizer and adaptiveregularization with quadratic line-search was implemented for gradient optimization. As theforward modelling and measurement errors for atmospheric inverse problems are usuallyunknown, issues concerning ‘model-fit’ and ‘data-fit’ were examined. In this paper, the