Strengthening
S > upper threshold
- Peaks
- —
- Mean phase
- —
- R
- —
- p-value
- —
Chinese speleothem record 0–640 ka
Monsoon / Changes
Thresholds calibrated over 20–60 ka
The charts could not be loaded.
The record
Drag to zoom · Click legend to toggle
Preparing the record…
▲ Strengthening ▼ Weakening
Age in ka before 1950 · Older → younger
Precession phase
All detected peaks · Rayleigh test
Weakening
S < lower threshold
- Peaks
- —
- Mean phase
- —
- R
- —
- p-value
- —
Radius = peak count · Line = mean resultant vector · 0° = precession minimum, 180° = maximum
Method & dataDefinitions, sources & limitations
From record to transitions
δ¹⁸O is interpolated to a 20 yr grid and smoothed with a 20 yr Gaussian. Within each window, S is the difference between the means of the upper and lower 5% of −δ¹⁸O, signed by their age order. Positive S indicates strengthening; negative S indicates weakening.
Peaks exceed the selected quantiles of S over 20–60 ka. Same-direction peaks are separated by at least 1 kyr; values are quantised to 10⁻⁴ ‰ before selection. The sample-gap strip shows distance to the nearest raw sample; its dashed line marks the 95th percentile.
Phase & interpretation
Phase interpolates between successive precession extrema: minimum = 0°, next maximum toward younger ages = 180°. Each histogram uses 30° bins. The mean vector has length R × the largest bin count, where R is the mean resultant length.
Interpret p-values cautiously. Overlapping windows and the record’s precession signal make peaks dependent; Rayleigh p-values may overstate evidence for clustering. Zooming changes the view, not the analysis interval.
Data sources
Chinese speleothem δ¹⁸O: Cheng et al. (2016). Precession: Laskar et al. (2004). Greenland GI/GS onsets: Rasmussen et al. (2014), with GICC05 ages × 1.0063.
Both δ¹⁸O and precession axes are reversed. Greenland onset markers are available through the legend and hidden initially to keep the full record readable. Interpolated gaps remain included in the analysis.