Hierarchical Dynamic Time Warping methodology for aggregating multiple geological time series

Burstyn, Yuval; Gazit, Asaf; Dvir, Omri

doi:10.1016/j.cageo.2021.104704

Cited by 9 publications

(2 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Dynamic time warping maximizes the similarity between two time series by warping, stretching, or compressing one sequence in the time dimension to match the other time series (Müller, 2007;Sakoe & Chiba, 1978) and has been successfully applied to provide robust age models for palaeo-records in other contexts (e.g. Burstyn et al, 2021). We adjust the reconstructed mean annual temperature binned into 5.625° × 5.625° grid cells, using the LOVECLIM simulated annual mean temperature (which includes the D-O cycles) as the reference.…”

Section: Dynamic Time Warpingmentioning

confidence: 99%

Rapid ice-age warming events amplified by strong vegetation-albedo feedback

Liu,

Prentice,

Menviel

et al. 2024

Preprint

View full text Add to dashboard Cite

The last glacial period was punctuated by >20 Dansgaard-Oeschger (D-O) cycles each manifested as a large, rapid (decadal-scale) warming in Greenland followed by slower cooling. D-O warming events have been identified as instances of a global climate-tipping point being crossed repeatedly. The mechanisms are not fully resolved. A role for the interaction of ice-sheet dynamics with the ocean circulation is widely accepted. Less attention has been paid to land-surface changes, although both decreasing snow and increasing tree cover in high latitudes could amplify warming. Here we use a global dataset of pollen records spanning multiple D-O cycles during Marine Isotope Stage 3 to reconstruct changes in land temperature, vegetation cover and vegetation height, and combine these with snow cover changes generated by the LOVECLIM climate model, using a 3D-variational data assimilation technique to estimate global patterns of change in land-surface albedo and temperature. We find a large (1.94 ± 0.65 W m–2 K–1) positive feedback from the combined effect of vegetation and snow, which is about 40 % larger than the pure snow-albedo feedback. This may be large enough to have destabilized the climate system, helping to explain the abruptness of D-O warming events.

show abstract

Section: Dynamic Time Warpingmentioning

confidence: 99%

Rapid ice-age warming events amplified by strong vegetation-albedo feedback

Liu,

Prentice,

Menviel

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…Although the ACER database provides age models for each pollen record, the resolution of the individual records is variable (mean resolution 474 years) and these models are often imperfectly aligned with the dating of D-O events as recorded in the Greenland ice core. To create a better alignment, we used dynamic time warping (DTW: Belman and Kalaba, 1959;Burstyn et al, 2021) to adjust the age scale for each individual record (Figure 2). Dynamic time warping optimises the similarity between two sequences by stretching or compressing one sequence in the time dimension to match the other.…”

Section: Age Modellingmentioning

confidence: 99%