[M-IS12] Paleoclimatology and paleoceanography

[MIS12-P05] Monthly to seasonal resolution time-series interpolation of paleoenvironmental records: accuracy and precision based on Tridacna shells

*Tatsuki Nakazawa¹, Hideko Takayanagi¹, Ryuji Asami¹, Jun Muto¹, Yasufumi Iryu¹ (1.Institute of Geology and Paleontology, Graduate School of Science, Tohoku University)

Keywords:A new time-series interpolation technique, Dynamic Time Warping, Giant clams, Oxygen isotope, Daily growth bands

In recent years, seasonality modulation and frequent extreme weather events worldwide have caused serious ecological disturbance and social problems such as food shortages and natural disasters. Some of these changes are believed to result from rapid anthropogenic climate change. However, the actual causes still need to be fully understood. One reason for the lack of clarity is that paleoenvironmental records, originally suitable for analyzing natural variations at monthly to seasonal resolutions dating back to a period without instrumental records, have yet to be analyzed at sufficient resolutions. For example, geochemical records of corals and giant clams provide primary data at monthly or higher resolution. However, in many cases, time-series interpolation of primary data commonly used in present studies results in significant temporal errors, preventing time-series analysis at a seasonal or higher time resolution. Therefore, it is essential to establish a new time-series interpolation technique that accounts for differences in time resolution of samples due to changes in growth rates to perform time-series analysis at a seasonal resolution or higher, dating back to preindustrial periods. In this study, we focus on the Dynamic Time Warping (DTW) method, which is suitable for data with such differences in time resolution of samples, and evaluate the accuracy and precision of this method for time-series interpolation of geological biogenic carbonate samples (such as corals and giant clams).
We performed growth line analysis and carbon (δ¹³C) and oxygen (δ¹⁸O) isotope measurements of two giant clams, Hippopus hippopus, collected from Sekisei Reef Lagoon between Ishigaki and Iriomote islands. The shells possess distinct daily growth bands, allowing for a direct comparison between the time determined by the daily growth bands and the time obtained by conventional and new time-series interpolation. The comparison indicated that the conventional time-series interpolation resulted in a time discrepancy of up to approximately one month, increasing the error as the distance from the anchor point (e.g., the annual maximum and minimum values) increased. On the other hand, the DTW-based times were consistent well with those obtained from the growth-line analysis, indicating that the DTW time-series interpolation is more suitable for geological samples than the conventional time-series interpolation. In this presentation, we present the accuracy and precision of the DTW time-series interpolation and discuss its potential as a new monthly to seasonal resolution time-series interpolation method.