| 07 Jul 2026
Assessing alignment-based chronology tuning for a 2000-year record of warm-water variability in Qeqertarsuup Tunua (Disko Bugt), West Greenland
Abstract. Chronological uncertainty limits the use of marine sediment cores for resolving the timing and coherence of late-Holocene ocean variability around Greenland. In this study, we apply two optimization methods to improve the chronological uncertainty of four sediment cores from Qeqertarsuup Tunua (Disko Bugt), West Greenland. Using the GISP2 temperature reconstruction as a reference series, we generate alignment-consistent age-model realizations with (i) Nelder-Mead optimization (NMO), which maximizes a resampled Pearson correlation coefficient (RPCC) and (ii) Dynamic Time Warping (DTW), which aligns sequential structure through non-linear time-axis adjustment. For individual cores, NMO generally yields higher RPCC values than DTW. However, when the adjusted records are merged into a Qeqertarsuup Tunua composite, DTW produces a substantially more coherent regional signal and a higher RPCC with GISP2 (0.77) than the NMO-based composite (0.36). To assess whether these improvements reflect genuine climate co-variability, we apply a phase-randomisation permutation test using 500 spectrally matched surrogate GISP2 series. Results are heterogeneous: DTW alignment is statistically significant for one core (Sullorsuaq, p=0.006), and NMO alignment is significant for another (Aasiaat1, p=0.034), while neither method reaches significance for the other two cores. Restricting the composite to these two significant core and method combinations yields an RPCC of 0.83 with GISP2, outperforming both other composites. Applying both methods to a synthetic reference series also increases RPCC values, demonstrating that correlation gains can arise from alignment flexibility alone and underscoring the need for independent chronological constraints when interpreting tuned chronologies. Mean age offsets relative to the original models range from -99 to 41 yr for NMO and from -346 to 190 yr for DTW. We conclude that NMO can be useful for site-specific alignment and that the significance-filtered composite provides the most defensible regional reconstruction of Atlantic Water variability over the past 2000 years.
In this manuscript, van der Laan and co-authors evaluate alignment methods based on the variability of marine sediment core data and warming/cooling transitions in temperature reconstructions to refine age-models and reduce apparent offsets among sediment cores at the regional scale. Specifically, they use published foraminiferal concentration records from marine sediment cores as a proxy for subsurface temperature and compare them with temperature reconstructions derived from the Greenland ice-core record. Both the Nelder-Mead Optimization (NMO) and Dynamic Time Warping (DTW) methods show potential for refining short-term (<2000 years) age models. The authors demonstrate that NMO is more suitable for optimizing individual core records, whereas DTW provides better fits when multiple records are considered simultaneously and is therefore more appropriate for regional-scale syntheses. Nevertheless, the authors appropriately highlight that these optimization methods should be used as refinement tools rather than as replacements for precise age models.
Overall, this manuscript is clear, well-written, and presents methodological approaches of interest for paleoenvironmental reconstructions and the paleoceanography community. The objectives are clearly stated, and the conclusions are well supported by the results. I have provided more detailed comments below, which I hope the authors will find helpful and constructive. I believe that a revised manuscript would merit publication.
General comments:
1) The authors apply two alignment methods in this study, Nelder-Mead Optimization (NMO) and Dynamic Time Warping (DTW). However, the reasoning for selecting these two methods over other available alignment approaches is not sufficiently explained. The manuscript would benefit from a more detailed justification of why NMO and DTW were chosen and whether alternative alignment methods were considered or tested. If other methods were evaluated but not included, a brief discussion of their performance or the reasons for their exclusion would strengthen the methodological framework.
2) The manuscript would benefit from a more detailed discussion of the potential risk of overfitting the marine sediment records to the Greenland ice-core record. Because the ocean subsurface responds to atmospheric temperature changes over timescales ranging from years to centuries, a direct alignment between marine and ice-core records may not always be justified. Optimizing the alignment could potentially bias the resulting age models by artificially shifting the marine sediment chronology, therefore increasing chronological uncertainties rather than reducing them. The authors should discuss this limitation more explicitly and clarify how their optimization procedures account for potential lags in the climate responses of the two records.
3) The evaluation of the proposed alignment methods is based on only four sediment cores. While this dataset may be appropriate for demonstrating the methodology, it is unclear whether it is sufficient to support the broader conclusions regarding the performance and applicability of NMO and DTW. The authors should discuss the limitations associated with the relatively small number of records analyzed and clarify whether the observed performance of the two methods is expected to be robust across a wider range of sediment cores and paleoenvironmental settings.
Specific comments:
Line 51. In the phrase “Resampled Pearson Correlation Coefficient”, only Pearson should be capitalized in this context. If you choose to retain this capitalization, please ensure it is used consistently in the Abstract as well.
Line 61. Please remove “in other regions”, as it is redundant with the beginning of the sentence.
Line 71. Please refer to Fig. 1 at least once in section 2.1 Qeqertarsuup Tunua.
Line 84. The references should be listed in chronological order, consistent with the rest of the manuscript.
Line 123. What are the “two Perner cores”? Do you mean the “two Aasiaat cores”?
Line 129. Please use the full name of the method for the header: “Section 2.5. Resampled Pearson Correlation Coefficient” and consider adding the acronym in parentheses. The same recommendation applies to Sections 2.6 and 2.7.
Line 132. This sentence is duplicated. Please remove the repeated version.
Line 136. As in the introduction, only Pearson should be capitalized in the phrase “Resampled Pearson Correlation Coefficient”.
Line 186. The acronym should be introduced by spelling out the full term first, followed by the acronym in parentheses.
Line 208. Figure citations should appear in sequential order. Fig. 7 is cited before Figs. 3-6. Please either revise the order of the figure citations or renumber the figures accordingly.
Table 3. In the caption, post-optimization should include a hyphen, and the unnecessary hyphen following core should be removed.
Figure captions. Please use the acronyms consistently throughout the figure captions, as done elsewhere in the manuscript. Some captions use the acronyms, whereas others use the full names of the methods.