Data Normalization Strategies for Full-waveform Inversion

Full-waveform inversion (FWI) relies on accurate estimation of signal amplitudes in order to produce reliable and practical results. Approximations made during the inversion process, as well as unknown or incomplete information, cause amplitude mismatch effects at various levels: shot gather absolute magnitude, trace-by-trace scaling and time-varying relative amplitudes. Rewriting the problem in an amplitude free formulation allows to mitigate the amplitude ambiguity and improve successful convergence. We present strategies to eliminate shot-gather and trace-by-trace amplitude uncertainty; we derive update directions, compare and analyze the effect of each of the normalization schemes.