1. 中国石油大学(北京)油气资源与探测国家重点实验室, 北京 102249;
2. 中国石油大学(北京CNPC:物探重点实验室, 北京 102249;
3. Department oI Earth Sciences,Uppsala University,Uppsala,Sweden;
4. 中海石油研究总院开发研究院, 北京 100027
Variable grid spacing velocity tomography
Huang Guangnan1,2, Liu Yang1,2, Ari Tryggvason3, Hu Guangyi4, Fan Tingen4, Dong Jianhua4
1. State Key Laboratory of Petroleum Resources and Prospecting,China University of Petroleum(Beijing),Beijing 102249,China;
2. CNPC Key Laboratory of Geophysical Exploration,China University of Petroleum(Beijing),Beijing 102249,China;
3. Department of Earth Sciences,Uppsala University,Uppsala,Sweden;
4. CNOOC Research Institute,Beijing 100027,China
Abstract：In variable grid spacing tomography theun-derground velocity distribution is parameterizedwith model cells of different sizes.This methodcan simultaneously take into account the spatiallyvarying resolution inherent in most datasets.Dueto experimental design or logistic constraints，theshallow and deep subsurface velocity distributionmay be very differently determined by the availabledata .The variable grid spacing tomography andregular grid spacing tomography are similar inmost other aspects.There are two main differencesbetween the variable grid spacing method and theregular grid spacing method.First，ray segmentscalculated in the forward model cells should be di-vided into the larger cells of the inversion model.Second，the smoothness constraint equations nee-ded to inhibit wild velocity variations need to bemodified where the cells change in size.In a syn-thetic checkerboard reconstruction test with differ-ently sized checkers，the variable grid spacingmethod recovers the small and large checkers bet-ter than the regular grid spacing method in bothtwo and three-dimensional test cases. For a realdataset，the variable grid spacing method distin-guishes two low velocity zones better than the reg-ular grid spacing method.Finally，it can be con-chided that when the effort is spent to identify there gions in a model with best data coverage.The variable grid spacing method can produce velocity images with higher fidelity than when a uniformcell size is used.Especially，in many cases thismethod can enhance the fidelity of the shallow sub-surface velocity distribution. In addition, vari ablegrid spacing tomography can reduce the underde-termined regions in an inverse problem when the data coverage is irregular by adjusting quantitative proportion of cells with different sizes.
