Polarization filter based on linear Radon transform to extract surface waves in 3C seismic data

Ivan Sanchez, William M. Agudelo, Daniel O. Trad, Daniel Sierra

Polarization filters are an alternative tool to attenuate ground roll when multicomponent seismic data are available. These filters can separate surface and body waves by discriminating elliptical from linear motions. Existing polarization filters work in single-station. Thus, they can filter seismic data with irregular spatial sampling. However, in most cases, polarization filters do not get results similar to $f-k$ filters because they do not operate in multitrace, and there is no velocity discrimination. We present a new polarization filter to extract surface waves from multitrace 3C seismic data. The filter works in the frequency-slowness domain as a mask on elliptic elements obtained from the linear Radon model of the seismic data in the frequency domain. We applied this filter on a 3C real shot gather from land seismic data acquired in the Middle Magdalena Valley in Colombia. The results show that this polarization filter can accurately extract the surface waves, preserving more low-frequency energy than the f-k filter.