基于非规则网格声波正演的时间域全波形反演

全波形反演是地震资料处理中速度建模的有力工具，相比层析成像等速度建模方法它能够得到速度场的更高频成分.本文给出了基于声波方程格子法正演的时间域全波形反演方法，该方法用非规则、非结构化的三角网格来离散计算区域及模型参数，能实现网格粒度与反演分辨率在空间上的自动匹配，内存需求少，计算效率高；采用L-BFGS优化方法，以分频段变网格的方式实施多尺度反演.以二维Overthrust模型进行了速度反演数值测试，显示了该方法的高效性和潜力.

Time-domain full waveform inversion based on an irregular-grid acoustic modeling method

Full waveform inversion(FWI) is a powerful tool for estimating the subsurface velocity model, which can achieve higher wavenumber components of the velocity model than conventional velocity modeling methods, such as traveltime tomography. We propose a time-domain full waveform inversion method based on acoustic modeling with the grid method, which uses irregular, unstructured grid to discretize the velocity model and parameter space. The grid scale can automatically match the resolution of inversion in model space, the acoustic modeling with irregular grid needs less memory and is more efficient than with regular grid. Combining multiscale strategy and adaptive irregular grid, we carry out FWI with L-BFGS as the optimizing method. This method is applied successfully to the 2D Overthrust model, of which the low and high wavenumber components are accurately obtained, which demonstrates its efficiency and potential.