oa Wave-Equation Propagation as a Body-Wave Filter

Passive seismic data collected with broadband seismometers is usually under-sampled. The sampling<br>problem makes it impossible to utilize multi-dimensional filtering techniques to isolate various<br>propagation modes. We propose a chain of wave-equation imaging and simple signal processing in the<br>image domain as an effective method to mitigate deleterious noise effects and improve the ability to<br>extract body waves containing subsurface information.<br>We begin with the principles of time-reverse modeling: data are reversed in time and inserted into the<br>model domain as source functions. In this domain, data are better sampled, and after sufficient<br>propagation steps wave-front healing will allow events within the data to coalesce into approximately<br>correct coherent wave fronts. At each time-step in the elastic propagation, the wave field can be<br>filtered to remove unwanted energy. Also, the propagation itself will confine high ray parameter,<br>undesirable, energy to the shallow section of the model domain due to evanescence in a v(z) medium.<br>We also capitalize on being able to exactly separate the P- and S-wave components using simple<br>vector operators. To accumulate a measure of body-wave energy content as a function of surface<br>coordinate, we sum the depth axis of the image after windowing away the contaminated shallow section.