Drill-rig noise suppression using the Karhunen-Loéve transform for seismic-while-drilling experiment at Brukunga, South Australia

• Published in: Exploration geophysics (Melbourne) 2016-03, Vol.47 (1), p.44-57
• Main Authors: Sun, Baichun, Bóna, Andrej, Zhou, Binzhong, King, Andrew, Dupuis, Christian, Kepic, Anton
• Format: Article
• Language: English
• ISSN: 0812-3985; 1834-7533
• DOI: 10.1071/EG14086

Diamond-impregnated drill bits are known to be low energy vibration seismic sources. With the strong interference from the drill rig, it is difficult to obtain the drill-bit wavefield with a surface receiver array. To overcome the challenge of surface wave interference generated from the rig for seismic-while-drilling (SWD), we need to separate the rig- and bit-generated signals. To this end, we apply two wavefield separation methods, the Karhunen-Loeve (KL) transform and the f - k filter, and compare their performance. The applicability of these methods is based on the drill rig and drill bit having different spatial positions. While the drill-bit spatial position changes during the process of drilling, the drill rig remains stationary. This results in the source wavefields from the drill rig and the drill-bit having different characteristics, and allows us to separate and extract the drill-bit signal. We use a synthetic model to compare the KL transform and f - k filter. Both techniques are robust when the noise wavefield has consistent amplitude moveout. However, for changing amplitudes, such as the rig noise, which has an unrepeatable wavefield due to power amplitude variation, we show that the KL transform performs better in such situations. We also show the results of signal analysis of the SWD experiment data acquired from Brukunga, South Australia. We demonstrate the feasibility of the KL transform in separating the coherent noises from the stationary drill rig in a hard rock drilling environment, particularly emphasising the suppression of the surface and direct waves from the rig. The results show that drill-rig noise can be effectively suppressed in the correlation domain.