Find the question you need
Choose a route or read from the beginning. All teaching chapters stay open; the grouped index below is expandable.
Separate the two axes
Chapters 1–2. Distinguish a reconstructed value from the time assigned to it.
Work through the rates
Chapters 3–5. Calculate a transition and follow a stretched chronology.
Handle unequal stretching
Chapters 6–7. Compare local rates and propagate duration uncertainty.
Assess the evidence
Chapters 8–10. Check resolution, chronology and common questions.
Practise and continue
Chapters 11–12. Test changed cases and choose further reading.
Full chapter index · Practice and worked checks · Palaeoclimatology subject root
A reconstructed climate transition can keep the same amplitude while its estimated rate changes. The measurements have not necessarily changed. The chronology may assign a longer or shorter duration to the same part of the archive.
This guide follows that distinction through original fictional examples. It extends the Palaeoclimatology subject guide by treating the age model as part of the rate calculation. It does not estimate the speed of any real historical climate event.
Chapter index
Separate the two axes · Chapters 1–2
1. Separate the archive coordinate from time
2. State which way the clock runs
Work through the rates · Chapters 3–5
3. Work one transition under two chronologies
4. Follow the rate through the coordinate change
5. Distinguish a clock shift from a clock stretch
Handle unequal stretching · Chapters 6–7
6. Allow stretching to vary within the event
7. Carry uncertainty through the duration
Assess the evidence · Chapters 8–10
8. Respect hiatuses, sampling and archive smoothing
9. Audit a claim that one event changed faster
10. Questions that clarify chronology and rate
Practise and continue · Chapters 11–12
11. Practise with a changed chronology
12. Choose a connected climate-history route
CHAPTER 1 OF 12 · Separate the two axes
1. Separate the archive coordinate from time
A sediment core or cave deposit supplies positions at which material is measured. Write z for depth, increasing downward. A reconstruction might assign a temperature estimate T(z) to each sampled position. A separate chronology assigns an age a(z).
These two relationships answer different questions. T(z) estimates what the proxy implies about climate at that position. a(z) places the position in time. An inferred climate history T(a) uses both.
Changing only a(z) leaves the depth-domain values unchanged but relocates them on the time axis. That can change durations and rates. It does not, by itself, supply new evidence for a different temperature calibration.
Here a is calendar age in years before 1950, abbreviated calendar years BP. Larger ages are older. Forward elapsed time increases in the opposite direction: t = C − a, where C fixes an arbitrary origin.
For a warming from an older value Told to a younger value Tyoung, the positive elapsed duration is D = aold − ayoung. The mean forward-time warming rate is (Tyoung − Told) / D. Using ayoung − aold in the denominator would reverse the sign.
One kyr means 1,000 years. A rate of 0.0015°C per year is 1.5°C per kyr. Attach the averaging interval as well as the unit: a centuries-long average does not claim a constant annual rate throughout that interval.
CHAPTER 3 OF 12 · Work through the rates
3. Work one transition under two chronologies
Imagine two sampled horizons. The deeper horizon at z = 3 cm has reconstructed temperature 10.0°C. The shallower horizon at z = 1 cm has 10.6°C. Assume, solely for this exercise, that the values represent comparable climatic quantities without additional calibration changes.
Chronology A assigns ages 1,500 and 1,100 calendar years BP to those deeper and shallower horizons. Chronology B assigns 1,700 and 1,100 years BP. Both preserve the deeper-before-shallower order.
| Quantity | Chronology A | Chronology B |
|---|---|---|
| Older horizon | 3 cm; 10.0°C; 1,500 years BP | 3 cm; 10.0°C; 1,700 years BP |
| Younger horizon | 1 cm; 10.6°C; 1,100 years BP | 1 cm; 10.6°C; 1,100 years BP |
| Reconstructed rise | 0.6°C | 0.6°C |
| Elapsed duration | 400 years | 600 years |
| Mean warming rate | 1.5°C per kyr | 1.0°C per kyr |
The numerical observations here are invented. The two chronologies are alternative teaching assignments, not equally probable models of a real archive. Actual constraints would determine which assignments are admissible and how much weight they receive.
CHAPTER 4 OF 12 · Work through the rates
4. Follow the rate through the coordinate change
Over a differentiable, continuous part of the record, the chain rule gives dT/da = (dT/dz) / (da/dz). Because forward time runs opposite to age BP, dT/dt = −(dT/dz) / (da/dz).
For a linear interpolation through the example’s two horizons, dT/dz = (10.0 − 10.6) / (3 − 1) = −0.3°C/cm. Chronology A has da/dz = (1,500 − 1,100) / 2 = 200 years/cm. Thus dT/dt = 0.0015°C/year.
Chronology B has da/dz = 300 years/cm and gives 0.0010°C/year. The same depth-domain temperature gradient is spread across more modelled time. These local derivative values rely on the stipulated linear interpolation; two endpoints alone do not demonstrate the actual rate at every moment between them.
CHAPTER 5 OF 12 · Work through the rates
5. Distinguish a clock shift from a clock stretch
An additive shift a′ = a + c moves every age by the same amount. Within that record, the duration between two fixed horizons is unchanged, so their mean rate is unchanged. Their placement relative to an independently dated event may change.
A uniform stretch around an anchor a0 has a′ = a0 + s(a − a0), with s positive. Durations multiply by s. For unchanged climate values and fixed horizon choices, mean rates divide by s. With a0 = 1,100 and s = 1.5, chronology A becomes chronology B over the example interval.
A common date shift and uncertain interval stretching therefore have different consequences. Better absolute timing does not automatically mean a better constrained transition rate. Ask which chronological component changes the duration used in the denominator.
CHAPTER 6 OF 12 · Handle unequal stretching
6. Allow stretching to vary within the event
Now add a middle horizon at 2 cm with reconstructed temperature 10.3°C. Consider chronologies C and D, each spanning 600 years from 3 cm to 1 cm. The middle age differs.
| Forward-time segment | Chronology C | Chronology D |
|---|---|---|
| Older 10.0°C to middle 10.3°C | 1,700 to 1,400 BP; 300 years; 1.0°C/kyr | 1,700 to 1,550 BP; 150 years; 2.0°C/kyr |
| Middle 10.3°C to younger 10.6°C | 1,400 to 1,100 BP; 300 years; 1.0°C/kyr | 1,550 to 1,100 BP; 450 years; about 0.667°C/kyr |
| Entire 0.6°C rise | 600 years; 1.0°C/kyr | 600 years; 1.0°C/kyr |
The whole-event mean rate is identical, but the location of faster apparent change differs. One global correction factor cannot represent a locally changing age-depth relationship.
The whole-event rate equals the duration-weighted mean of the segment rates. Averaging D’s two rates without their unequal durations would give the wrong whole-event result. The reconstructed temperature change is recovered by adding each rate multiplied by its own duration.
CHAPTER 7 OF 12 · Handle unequal stretching
7. Carry uncertainty through the duration
Hold the 0.6°C amplitude fixed for a simple sensitivity test. A duration of 300 years gives 2.0°C/kyr; 500 years gives 1.2°C/kyr. Relative to the 400-year result of 1.5°C/kyr, the two rate changes are unequal because rate depends on the reciprocal of duration.
Those two alternatives are a sensitivity calculation, not a confidence interval. A probability statement needs an actual uncertainty model. Real reconstructions may have uncertainty in both the climate values and the chronology, with dependencies among samples and dates.
For a chronology ensemble, calculate the defined duration and rate within each valid member, preserving its paired ages and climate inputs. Summarise those derived rates appropriately. Independently combining an oldest possible endpoint with a youngest possible endpoint can create an interval unsupported by any admissible chronology.
CHAPTER 8 OF 12 · Assess the evidence
8. Respect hiatuses, sampling and archive smoothing
A hiatus can place substantial elapsed time across little or no deposited material. A smooth derivative formula should not be pushed through that discontinuity as though continuous accumulation had been observed. An interval-average rate across the gap remains a statement about its endpoints, not a reconstruction of the missing trajectory.
Likewise, making an interpolated curve annually spaced does not create annual observations. Sampling, mixing and proxy response can suppress a short rapid transition. A very steep interpolation between two noisy, closely assigned ages does not independently establish an extreme climate-change rate.
The COPRA primary methods paper treats age modelling, reversals, hiatuses and propagation of dating uncertainty into proxy records. Werner and Tingley’s age-depth reconstruction study develops a probabilistic approach to time-uncertain proxies. Our simple coordinate calculations illustrate the issue; they do not reproduce those research models or their results.
CHAPTER 9 OF 12 · Assess the evidence
9. Audit a claim that one event changed faster
Ask which climate quantity was reconstructed, how its amplitude was calibrated, which horizons define the transition, and which dates constrain their separation. Preserve the age convention and the exact duration used.
Then compare compatible windows. A maximum rate over a short interval and a mean rate over a long one are different statistics, even with perfect dates. Establish whether archives have comparable effective resolution and whether smoothing, sampling or event selection changed the apparent maximum.
Finally, ask whether an alignment assumption helped build the chronology. If a curve was stretched to match a presumed event duration, that duration cannot also serve as independent confirmation of the resulting rate. Make the role of each constraint visible.
CHAPTER 10 OF 12 · Assess the evidence
10. Questions that clarify chronology and rate
Does an uncertain rate mean the warming amplitude is unknown?
Not necessarily. This article deliberately holds amplitude fixed to isolate duration. In real work, amplitude has its own calibration and sampling uncertainties. State both rather than allowing one uncertainty to stand for every part of the reconstruction.
Can every age model be stretched freely?
No. Dating evidence, stratigraphic order, layer counts, accumulation assumptions and hiatus information constrain permissible chronologies. A mathematical transformation is not automatically a scientifically admissible age model.
Does a larger age gradient mean faster climate change?
For an unchanged temperature gradient per depth, a larger years-per-centimetre gradient means more elapsed time across that change and therefore a smaller rate magnitude. Keep the denominator’s physical meaning in view.
Case A: a fictional 0.8°C rise occurs between horizons aged 2,400 and 2,000 BP. The mean warming rate is 0.8 / 400 × 1,000 = 2.0°C/kyr. Add 200 years to both ages. The rate stays 2.0°C/kyr because the duration remains 400 years.
Case B: keep the younger age at 2,000 BP but move the older horizon to 2,800 BP. Duration is now 800 years and the rate is 1.0°C/kyr. Name what changed: the chronology’s interval length, with the reconstructed amplitude held fixed.
Case C: a report offers 0.3°C over 100 years beside 0.9°C over 1,000 years and calls the latter faster because its amplitude is larger. The mean rates are 3.0 and 0.9°C/kyr respectively. Before treating that comparison as a real climate conclusion, also check calibration, uncertainty and effective temporal resolution. Correct arithmetic is one part of a defensible comparison.
CHAPTER 12 OF 12 · Practise and continue
12. Choose a connected climate-history route
Reconstruct the archive and its clock
Return to Palaeoclimatology and the broader paleoclimate evidence guide. Use Geochronology for numerical dates and Stratigraphy for sequence and missing intervals.
Distinguish related sources of apparent change
Shared age anchors and lead–lag uncertainty keeps chronological dependence intact when comparing timing. Sampling intervals and aliasing explains false periodic patterns. Seasonal proxy weighting separates a proxy signal from the climatic average it may be taken to represent.
Transfer the reasoning
The Gold Standard of Education and How X Works connect the subject to wider learning. Whenever a claim concerns speed, identify how the clock was constructed as carefully as the change being timed.
Sources checked 21 October 2026. Worked examples are educational models with the assumptions stated above.