Ocean Turning Points

Science · Methods

How we handle evidence

The aim is that anyone can see where a number comes from, what kind of evidence it is, and how much weight it can bear.

Sources

We use peer-reviewed research, assessments by the Intergovernmental Panel on Climate Change (IPCC), and official, openly licensed datasets. Each reference is checked against its publisher’s record before it is cited. We do not publish new research or unpublished results, and we do not use material shared in confidence.

Kinds of evidence

Every map and chart is labelled with the kind of evidence it shows, because they answer different questions.

  • Observed Direct measurements, such as the RAPID array. They tell us what has happened, over the period measured.
  • Model experiment Idealised simulations designed to isolate a process, such as adding fresh water until the AMOC collapses. They show the effect of a change. They do not say when, or whether, it will happen.
  • Forecast Seasonal outlooks issued by forecasting centres such as NOAA. Their skill is measured and falls with lead time.
  • Projection Simulations driven by a scenario of future emissions. They are conditional on that scenario and are not forecasts.

We also keep apart historical and future periods; a scenario and a prediction; exposure (who or what is in harm’s way) and damage (what is lost); and a hypothetical comparison, such as a coast with no reef at all, and a realistic one, such as a reef that degrades.

Uncertainty and confidence

Where the IPCC has assessed a statement, we use its calibrated language: confidence (low, medium, high) for the strength of evidence and agreement, and likelihood terms (likely, very likely) for probabilities. Where it has not, the “How sure are we?” tables give our own reading of the published evidence, with the basis stated.

When several models are available we show the range across them as well as the average. Rainfall maps mark where all models agree on the direction of change, and tables list each model’s value.

Processing

All processing is done by a documented, open pipeline that downloads the original files, records their sources and checks results against the published papers. Model fields are placed on a common 0.5° latitude-longitude grid, finer than any model used, so no model detail is lost; the finer grid adds no real detail. The same decoded values drive the colours and the numbers shown when you hover, so what you see and what you can read are always identical.

Colour scales are perceptually uniform and readable with common forms of colour-blindness: cmocean “balance” for temperature (Thyng et al., 2016) and the brown-to-blue-green ColorBrewer scale for rainfall, the convention used by the IPCC. Each variable has a fixed, stated range that is shared across comparisons, so colours mean the same thing from one map to the next. Values beyond the range are shown at the end colour and marked “≤” or “≥” in the legend.

Values for named places are taken from the model grid cell containing the place. Climate-model cells are typically 100 to 250 km across, so these figures describe the surrounding region.

Limitations

  • The AMOC temperature maps come from a single model (CESM). Other models give the same pattern of change but different magnitudes.
  • AMOC experiments use artificial freshwater forcing to produce a collapse. Ice sheets are fixed, and greenhouse forcing is held constant after 2100.
  • Four models is a small ensemble. The average shows where models agree; it is not the single most likely outcome.
  • Rainfall changes are percentages relative to each model’s pre-industrial climate; they are not added to today’s observed rainfall.
  • Daily satellite products are less reliable near coasts and sea ice. Coral heat stress indicates bleaching risk; how reefs respond varies locally.

Review and corrections

Each topic shows the date it was last reviewed against new research. Topics are reviewed at least once a year and when major studies or assessments are published. Corrections are made openly and noted on the page concerned.

References

  1. Ben-Yami, M., Good, P., Jackson, L. C., et al. (2024). Impacts of AMOC collapse on monsoon rainfall: a multi-model comparison. Earth’s Future, 12, e2023EF003959. doi:10.1029/2023EF003959
  2. Jackson, L. C., Alastrué de Asenjo, E., Bellomo, K., et al. (2023). Understanding AMOC stability: the North Atlantic Hosing Model Intercomparison Project. Geoscientific Model Development, 16, 1975–1995. doi:10.5194/gmd-16-1975-2023
  3. Román, M. O., Wang, Z., Sun, Q., et al. (2018). NASA’s Black Marble nighttime lights product suite. Remote Sensing of Environment, 210, 113–143. doi:10.1016/j.rse.2018.03.017
  4. Thyng, K. M., Greene, C. A., Hetland, R. D., et al. (2016). True colors of oceanography: guidelines for effective and accurate colormap selection. Oceanography, 29, 9–13. doi:10.5670/oceanog.2016.66
  5. van Westen, R. M. & Baatsen, M. L. J. (2025). European temperature extremes under different AMOC scenarios in the Community Earth System Model. Geophysical Research Letters, 52, e2025GL114611. doi:10.1029/2025GL114611