Conclusions

Multi-component source equalization can be carried out in the time domain using short 1-D filters. Reciprocal experimental geometry is necessary for this method to work properly. In the data example above data acquisition is nearly reciprocal and receiver locations can easily be interpolated onto source locations. No subsurface information such as velocity, density, or stiffness is necessary for performing the equalization. That makes this method a preferred preprocessing step. The disadvantages are the potentially low signal-to-noise ratio and the implicit averaging over source emergence angles. The estimated filters mainly show amplitude variations and small modifications to the source-time function. My next step is to check the influence of this equalization on data rotation, stacks, and migrated images. I plan to implement a similar scheme that estimates absolute radiation patterns with an approximate knowledge of subsurface properties.