Abstract
The Hikurangi margin is one of the largest sources of seismic and tsunami hazards in New Zealand, but there is still much that remains unknown about previous ruptures on the subduction interface. Turbidite paleoseismology has the potential to increase the spatial density and temporal extent of paleoearthquake records. However, it is heavily reliant on temporal correlation of turbidites, and thus, requires them to be precisely dated. Typically, ages are obtained using radiocarbon dating of pelagic foraminifera from background sediments deposited between turbidites. This dating method requires background sedimentation to be accurately distinguished from the fine-grained tails of turbidites. Along the southern and central Hikurangi Margin, background sedimentation and turbidite tails have proven difficult to distinguish from one another. Here, a quantitative approach is developed to distinguish turbidite tails and background sediments using machine learning. <br><br>This study utilizes a natural experiment generated by the M W 7.8 Kaikōura earthquake, which caused the deposition of co-seismic turbidites at locations both proximal and distal to active canyon systems. The 2016 turbidite could be recognised due to its stratigraphic position at core tops. Turbidites and background sediments were independently identified using 210 Pb activity profiles to identify gradual accumulation. Additionally, foraminiferal assemblages were used to identify transported material. The physical and geochemical properties of the sediments were then analysed using non-destructive (computed tomography density, magnetic susceptibility, micro X-ray fluorescence derived geochemistry) and destructive (grain size, carbonate content, organic content) techniques, to develop a quantitative definition of turbidite tails and background sediments. The destructive datasets were then compared to the non-destructive data that acts as a proxy for these analyses because the latter are rapidly generated at high resolutions down core and are now routinely acquired in most turbidite paleoseismology studies. It was determined that there was a statistically significant correlation between the destructive data and the non-destructive proxies, such that the non-destructive data could be used as a viable alternative to the time consuming destructive analyses. <br><br>The machine learning technique, Linear Discriminant Analysis (LDA), successfully distinguishes background sediment and turbidite tails in areas where they are visually indistinguishable. The LDA model shows that in cores distal from active canyon systems, background sediment and turbidite tails are more distinct than in cores proximal to active canyon systems. Difference between canyon-proximal and distal sites may be due to the impact of weak bottom currents that are inferred to be acting on the background sedimentation processes along this margin. This study shows that quantitative identification of background sediments and turbidite tails is possible, and could allow for more robust identification and dating of turbidites globally, which is of paramount importance for the effective application of turbidite paleoseismology.<br>
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.