SVD and seismic interferometry for attenuation of ice-sheet flexural waves

In recent years seismic interferometry was extensively investigated and studied from theoretical point of view. Many applications using interferometry by crosscorrelation and by crossconvolution have been proposed to redatum the seismic wavefields, to extract new Green's functions, and to estimate and subtract noise wavefields. In this work we study an interferometry application with noisy data recorded on ice plates and affected by strong flexural noise. We perform the dispersion analysis and use a singular value decomposition (SVD) to extract the flexural noise from the eigenimages of the wavefields recorded in an Alaska on-ice seismic-reflection experiment. Similar filtering methods are used for attenuation of the ground roll. The SVD noise-estimation method gives improved results with respect to the original data, which are then combined using seismic interferometry to obtain redatumed sea-bottom signals in the shallow water layer below ice. The SVD results allow us to better identify the seismic reflections in the interferometry shots without need of surgical bandpass filters and to use improved signals for subsequent signal stacking and imaging.