Wavelets are critical to inversion methods. Incorrect phase estimation will affect the objective function and cause convergence to local minima, and thus produce biased or incorrect results. Based on two simple models and ignoring all other factors, we studied the variation of the wavelet phase as a function of frequency and its effect on the prestack waveform inversion. Numerical experiments show that an incorrect phase may result in large deviations from the real solution, even if there is a high similarity between the model and real wavelets. The precision of the inversion slightly improves by using the constant-phase rotation; however, the effect of phase inaccuracy is not eliminated, which limits the precision of prestack inversion.
