初至波波形反演方法及其数值模拟试验

First-arrival waveform inversion method and its numerical test

(1.同济大学海洋与地球科学学院，上海200092；2.中国卫星通信集团公司上海分公司，上海200233)

School of Ocean & Earth Science, Tongji University, Shanghai 200092, China

近地表速度模型的反演精度是影响复杂地表条件下地下地质构造成像效果的主要因素之一。初至波波形反演方法将速度建模问题转化为寻求地震数据信息和理论模型的最佳拟合，即转化为求解目标函数极小化问题，利用观测记录和模型正演记录之间的最佳匹配关系建立地下模型。梯度类波形反演方法的关键因素是目标函数、梯度和步长，梯度问题可以通过确定合适的目标函数来解决，选取合适的步长可以有效地提高波形反演的计算效率和反演精度。初至波波形反演方法类似于对地表模型同时做偏移和层析，能够同时分辨出高波数和低波数成分，并且同时恢复速度值和速度界面。初至波波形反演结果与初至波射线走时层析反演结果的对比分析表明，对于速度变化比较缓慢的起伏地表，前者的反演效果明显优于后者。

The inversion accuracy of near surface velocity model is one of the main factors for influencing the imaging of subsurface geologic structure with complex surface conditions. First arrival waveform inversion transforms velocity modeling problem into finding best fitting between seismic data information and theoretical model, which is to solve minimization problem of objective function, and use the best matching relationship between observation record and model forwarding record to establish subsurface model. The key factors for gradient-type waveform inversion method are objective function, gradient and step length. Gradient problem can be solved by defining proper objective function. Selecting proper step length can effectively improve the computation efficiency and inversion accuracy. First arrival waveform inversion method is similar to simultaneous migration and tomography on surface model, which can simultaneously recognize high wave number and low wave number component, and recover velocity value and interface. The first arrival waveform inversion results are compared to first arrival ray travel time tomography inversion results. The comparison results show that the former inversion is obviously superior to the later one at the rough surface with slow velocity change.