Vibroseis Acquisition Robust to Ambient Noise Using Pseudo-Random Sweeps and Noise-Aware Deconvolution

Non-stationary ambient noise is a pregnant issue when acquiring seismic in urban areas with a dense road traffic. It is usually handled with diversity stacking of repeated sweeps.

We propose here to handle this noise using single, pseudo-random sweeps at each station. Because all frequencies are emitted at all time, bursts of noise in the uncorrelated records can be weighted down and the earth refelectivity reconstructed from the cleaner parts of the record.

Instead of correlation, a noise-aware deconvolution is used. It relies on the recorded ground-force to shape the wavelet, and takes into account where the uncorrlated record has been weighting down.

This combination of pseudo-random signals and noise-aware deconvolution is demonstrated on synthetic data, as well as an actual 2D urban acquisition.