Neighbourhood algorithm and its application to 1D magnetotelluric data inversion
HUANG Wei-Hang1,2, JIN Wei-Jun1, ZHANG Wen-Hui1,2
1. Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing 100029, China;
2. University of Chinese Academy of Sciences, Beijing 100049, China
Magnetotelluric method (MT) is applied so widely that improving the method of inversion and the interpretation of MT data become very important. This paper summarizes different inversion algorithms and points out their limitations. Then an overview is given on neighborhood algorithm (NA) and its application to the MT inversion. The maximum likelihood model from NA is very close to the theoretical model. Although NA and genetic algorithm (GA) have the same capability for resisting the local minimum value, NA algorithm shows better similarity between the density distribution of sampling points and the error function, so NA is more conducive to inversion application based on the integral parameter appraisal method. In this study, the authors found that NA converges much faster than GA, whereas NA algorithm is more advantageous than GA algorithm in 1D MT inversion.
[1] Bedrosian P A. MT+, integrating magnetotellurics to determine earth structure, physical state, and processes[J]. Surveys in Geophysics, 2007, 28(2-3):121-167.
[2] Bostick F X. A simple almost exact method of MT analysis[C]//Workshop on electrical methods in geothermal exploration. 1977:175-177.
[3] 王家映, LevyS O D. 大地电磁测深的拟地震解释法[J]. 石油地球物理勘探, 1985, 20(1):66-79.
[4] Constable S C, Parker R L, Constable C G. Occam's inversion:A practical algorithm for generating smooth models from electromagnetic sounding data[J]. Geophysics, 1987, 52(3):289-300.
[5] Mackie R L, Madden T R. Conjugate direction relaxation solutions for 3-D magnetotelluric modeling[J]. Geophysics, 1993, 58(7):1052-1057.
[6] 陈小斌, 赵国泽, 汤吉, 等. 大地电磁自适应正则化反演算法[J]. 地球物理学报, 2005, 48(4):937-946.
[7] 陈孝雄, 王友胜. 利用初始模型提高 MT 反演分辨率[J]. 江汉石油科技, 2009, 16(4):19-22.
[8] Hammersley J M, Handscomb D C. The General Nature of Monte Carlo Methods[M]. Springer Netherlands, 1964.
[9] Rothman D H. Nonlinear inversion, statistical mechanics, and residual statics estimation[J]. Geophysics, 1985, 50(12):2784-2796.
[10] Spichak V, Popova I. Artificial neural network inversion of magnetotelluric data in terms of three-dimensional earth macroparameters[J]. Geophysical Journal International, 2000, 142(1):15-26.
[11] Stoffa P L, Sen M K. Nonlinear multiparameter optimization using genetic algorithms:Inversion of plane-wave seismograms[J]. Geophysics, 1991, 56(11):1794-1810.
[12] Shaw R, Srivastava S. Particle swarm optimization:A new tool to invert geophysical data[J]. Geophysics, 2007, 72(2):F75-F83.
[13] Sambridge M. Geophysical inversion with a neighbourhood algorithm-I. Searching a parameter space[J]. Geophysical Journal International, 1999, 138(2):479-494.
[14] Voronoï G. Nouvelles applications des paramètres continus ã la théorie des formes quadratiques. Deuxième mémoire. Recherches sur les parallélloèdres primitifs[J].Journal für die reine und angewandte Mathematik, 1908, 134:198-287.
[15] Jones A G, Hutton R. A multi-station magnetotelluric study in southern Scotland-Ⅱ. Monte-Carlo inversion of the data and its geophysical and tectonic implications[J]. Geophysical Journal International, 1979, 56(2):351-368.
[16] Sambridge M. Geophysical inversion with a neighbourhood algorithm-Ⅱ. Appraising the ensemble[J]. Geophysical Journal International, 1999, 138(3):727-746.