the 22nd Pacific−Asian Marginal Seas meeting (PAMS)

Determining sedimentary depositional features in deep-marine environments using non-destructive techniques has become increasingly important for high-resolution paleoenvironmental reconstruction. In this study, we used two piston sediment cores (MD18-3532 and MD18-3533) recovered during the 2018 EAGER cruise from the Ryukyu accretionary wedge. MD18-3532 and MD18-3533 were retrieved from perched settings near the Nanao and Hoping basins, respectively, both of which are sub-basins of the Ryukyu forearc system. We integrate multiple non-destructive datasets, including color reflectance and physical property measurements obtained from a multi-sensor core logger (MSCL), computed tomography (CT) scan imagery, and X-ray fluorescence core scanning (XRF-CS) data. Multivariate statistical analysis of XRF-CS data has been widely applied in event stratigraphy and is increasingly used to identify depositional changes. In this study, the XRF-CS dataset was processed using centered log-ratio (clr) transformation followed by k-means clustering. This approach enables the detection of sedimentary signals associated with turbidite plume deposits that can overtop basin topography and reach perched basin settings. The results further reveal shifts in sediment provenance associated with the transition from glacial to Holocene periods. The primary objective of this study is to differentiate sedimentary signals related to high export productivity from those recorded in the background sedimentary sequence. However, our preliminary results indicate that clr-based clustering of XRF-CS data alone has limited sensitivity in resolving biogenic versus detrital variability over time. To better identify intervals of elevated biogenic productivity, we apply z-score normalization to physical properties and spectrophotometric parameters prior to clustering. This approach enhances the detection of intervals associated with high organic export flux (mainly Laminated Diatom Mats (LDMs)). In addition, CT scan data provide complementary insights, with low-density regions corresponding to LDMs with high porosity. Overall, the preliminary results of this study demonstrate that the effectiveness of multivariate statistical techniques depends strongly on the type of input data. A combined approach integrating multiple non-destructive proxies with tailored statistical treatments significantly improves the identification of depositional processes and sedimentary features in complex accretionary wedge environments.
 

Ryukyu accretionary wedge area,XRF core scanning (XRF-CS),Multivariate analysis,Paleoenvironmental reconstruction,Laminated Diatom Mats