以声波时域全波形反演原理为基础,分析了数据完备性、数据噪声、正问题(包括正问题表征、震源子波、边界及频散等)、初始模型、正则化以及多尺度等因素对时域全波形反演的影响,并提出了相应的实用化建议;探索了加入井数据约束的反演策略,给出了相应的井约束全波形反演误差泛函与梯度计算公式以及CPU并行实现流程,利用复杂地震地质模型分析验证了方法的正确性.
Based on the theory of acoustic full waveform inversion(FWI) in time domain,we have analyzed the effects of factors such as data completness,data noise,modeling problems(modeling problem characterization,the source wavelet,boundary and dispersion),initial model,regularization and multi-scale problems on FWI in time domain,proposed the practical suggestions.And we have explored the inversion strategy with well-data constrained,researched the formula of error function and velocity gradient and implementation process of CPU parallel for well-constrained FWI in time domain.Finally,the correctness of this method is verified by the example of complex seismic geologic model.