| Popis: |
Summary We propose a dual frequency phase difference objective function specially designed for land data FWI. It is constructed by calculating the phase difference between two selected frequencies for both simulated and observed data respectively, then minimizing the difference between the phase differences. Comparing to other FWI objective functions, the advantages of our method are: first, by carefully select the interval between the two frequency, we can always keep the phase differences inside one cycle (2π) thus avoid cycle skipping; second, because it only use the phase information of the wave field, it has very high tolerance to unreliable amplitude in observed data; finally, this method is independent to source and receiver signature thus eliminating the need to invert them. These advantages especially last two are particularly important to land data FWI because: first, land data is generally much more noisy than marine data and its amplitude is more unreliable; second, it is common for land data that sources and receivers have their own signatures because of different coupling conditions, therefore a source/receiver signature independent method is a must for land data FWI. We demonstrate the effectiveness of our method using a model data set and two field data sets (Inner Mongolia 2D and Pre-Caspian basin 3D). |
