Upper left: the Marmousi model; Upper right: the "compressed" Marmousi model (17:1); Lower left: the MRA display of the model. The DC is overexposured to make the high frequency features visible. Lower right: the residual. Again, here the compression ratio does not have the meaning as in general data compression.
Since the Marmousi model is though to be complicated, it is thus safe for many 2-D inversions to use this compression ratio. This mainly the model size of the inversion is reduced by 17 times, the speed-up of the inversion can be far higher than this, because inverse problem computation is not of O(n).
Reference: 2-D Tomography in Multi-resolution Analysis Model Space by Z. Meng & J. Scales
Maintained by Zhaobo Meng