A robust source-independent full-waveform inversion

Xin Fu, Kristopher A. Innanen

Full-waveform inversion (FWI) can reconstruct high-resolution underground velocity and lithology structures even under complex geological backgrounds, and has been widely developed. But a reliable real-data inversion generally needs accurate source wavelet information, which is still one of the major challenges in FWI. In this paper, a robust source-independent FWI method is developed, which is demonstrated via synthetic tests of different starting models, different true models, different levels of random noises, and different types of source wavelets. It does not require any prior source wavelet information. It does not require an accurate starting model, even a 1D starting model is feasible to output an accurate wavelet estimate. It is stable for random noises. A good estimate of the source wavelet can be obtained from a poorly converged model based on the new proposed wavelet estimation equation. All in all, the performance of the new source-independent FWI in the synthetic data tests is close to that of the known-source-wavelet FWI.