This vibroseismic survey was conducted on Thwaites Glacier in West Antarctica in 2022/23, as part of the International Thwaites Glacier Collaboration's project GHOST (Geophysical Habitat of Subglacial Thwaites). The 210 km long seismic profile was measured parallel to the average flow direction. The processing of the recorded vibroseismic data to produce a Kirchhoff migrated section involves several standard steps outlined below. The recorded raw vibroseismic data were compressed by a cross-correlation with a synthetic source signal. The individual traces were assigned geometry from idealized shot locations and group intervals to organize common midpoint (CMP) gathers. Next, noisy or dead traces were removed. In particular, we removed the eight channels closest to the source as these were generally clipped. A frequency filter from 10–190 Hz was applied to reduce high-frequency noise, and a notch filter at 190 Hz was applied to reduce ringing from the spurious response of the geophones. We performed a zero-phase spike deconvolution to compress the wavelet to a spike. Afterwards, we re-sorted the data into CMP gathers and determined the stacking velocities. Stacking velocities were analysed using an automated constant velocity stacking method. We tested numerous NMO corrections for a broad range of velocities spanning from 3000 to 5000 m/s with 10 m/s intervals. We divided the TWT into segments of 20 ms with 10 ms overlap. For each segment, we identified the optimal stacking velocity based on the stack with the highest amplitude. Next, we estimated the stacking velocities of the ice base, of the base of geological features beneath the ice and of the bed. The stacking velocity at the ice base was calculated from the average of three TWT segments encompassing the ice base. A similar procedure was applied for the geological features. To determine the stacking velocity of the bed, we averaged the velocities of five TWT segments situated approximately 250 ms beneath the ice base or beneath geological features, respectively. For ice and bed, we averaged the stacking velocities over a wide range of 500 CMP gathers and for geological features over 50 CMP gathers. The resulting velocity field was used to perform the normal moveout (NMO) correction. We stacked the NMO-corrected CMP gathers to improve the signal-to-noise ratio. To remove diffraction hyperbolas, we migrated the stacked section with a time-space Kirchhoff migration using interval velocities of ice, geological features and the bed. We estimated the interval velocity of ice to be 3770±35 m/s from an alignment of the ice thickness with airborne radar-derived ice thicknesses. The interval velocities of geological features were determined based on the Dix-Dürbaum-Krey equation using the stacking velocities. We estimated the interval velocity for a large sedimentary basin to be 2680±440 m/s. We assumed a constant interval velocity of the bed of 4000 m/s. Finally, we applied a two-way traveltime (TWT) to depth conversion with these interval velocities. This dataset contains the Kirchhoff migrated section in standard SEG-Y format. The individual traces were assigned idealized geometry of CMP locations. The CMP coordinates, bed elevation, ice thickness, and basal slope are given in the .csv file.

If you have questions, please contact Ole Zeising (ole.zeising@awi.de) or co-authors.Data were acquired in the field by: Ole Zeising, Coen Hofstede and Olaf Eisen.Data processing was done by Ole Zeising.This survey has been implemented by Alex Brisbourne and Sridhar Anandakrishnan as part of the ITGC-GHOST Project.The authors would like to thank Aspen Technology, Inc. for providing software licenses and support.