The time travel: from today back to 540 million years
What was the climate like in the past? Here temperature can be compared at six scales, together with the Vostok ice core, the reconstruction of the last 2,000 years and the question of whether warming is faster today than before.
How to read: Measurement measured values · Reconstruction from indirect data such as ice or tree rings · Model computer model · Own calculation calculated by me from raw data. Technical terms are explained in the glossary.
From today back to the ice age
One chart, six scales: from the last 90 years in Amstetten to 540 million years of Earth history. Start the time travel or drag the slider, and move the mouse or a finger over the curve. The golden window shows how far we saw in the previous step.
With the "Period" slider below the chart you can move continuously through the displayed period, from the last 90 years to 540 million years.
Sources and data of the time travel: Amstetten station: GeoSphere Austria and own measurement. 2,000 years: Ljungqvist 2010. Vostok ice core (temperature and CO₂): Petit et al. 1999. CO₂ of the last 2,000 years: MacFarling Meure et al. 2006 (Law Dome) and NOAA Mauna Loa. Up to 66 million years: CenCO2PIP 2023 (CO₂, data) and Westerhold et al. 2020 (temperature). Older: Scotese et al. 2021 (temperature, data) and Foster et al. 2017 (CO₂). Transitions: the series are not aligned to each other at dataset changes (except ice core and reconstruction over the common period of the last 2,000 years, by mean offset); changes are marked in the chart.
💬 In plain language
- Old weather is locked in the ice of Antarctica. Researchers have drilled out a very long piece of ice, the Vostok ice core. It reaches back 420,000 years.
- Tiny air bubbles are trapped in the ice. From them we know how much CO₂ was in the air back then. The temperature is derived from the water in the ice.
- The blue line is the temperature, the golden one is the CO₂. Both rise and fall together across all four ice ages. So they are closely linked.
- On average, the CO₂ comes somewhat after the warmth, about 1,000 years later. However, the age of the ice can only be determined to within a few hundred to a thousand years.
- The temperature applies to Antarctica and not to the whole Earth.
Temperature and CO₂ over four ice-age cycles
What it shows: The Vostok ice core in Antarctica reaches back about 420,000 years. Temperature change (blue, from the deuterium in the ice, relative to today) and the CO₂ content of the trapped air (gold, ppmv) are overlaid in one chart. Time runs from left (420,000 years ago) to right (today).
What you see: Both curves rise and fall together over all four ice-age cycles (correlation 0.86). With the slider the CO₂ curve can be shifted back in time; the agreement is greatest at about 1,000 years: on average the CO₂ therefore lags behind the temperature. The dating of ice and air is accurate only to within a few hundred to a thousand years, and the offset is of this order of magnitude. The correlation and the “best” offset come from a search over many offsets (−3,000 to +6,000 years); this is an exploratory analysis and not proof of a cause-and-effect relationship. That CO₂ lags on average fits a feedback (a warming triggered in another way releases CO₂, which then warms further) and says nothing about whether additional CO₂ warms today.
On scope: The temperature is a value derived from the ice (deuterium) for Antarctica, not a global mean.
💬 In plain language
- Thermometers have only existed for about 170 years. How warm was it before that? We read this from traces in nature: from tree rings, from deposits on the sea floor and from ice cores.
- The blue line is reconstructed in this way; the golden one is measured with thermometers. Where both exist at the same time, they agree well.
- You can see that in Roman times and in the Middle Ages it was about as warm as 1961 to 1990, with colder periods in between. From about 1300 to 1900 came the “Little Ice Age”, coldest around 1700. So the temperature has always gone up and down.
- The curve applies to the Northern Hemisphere outside the tropics and not to the whole Earth.
Warm periods and cold phases since the birth of Christ
Blue: reconstruction from proxy series (tree rings, sediments, ice cores). Gold: thermometer measurements from 1850. Both show the same quantity in two ways and agree well up to 1989 (on average 0.05 °C, at most 0.10 °C difference). In the last decade, 1990–1999, the reconstruction is at +0.06 °C, the thermometer at +0.39 °C.
What it shows: The temperature of the Northern Hemisphere outside the tropics (30–90 °N) in decadal values as an anomaly from the 1961–1990 mean. The blue line is reconstructed from 30 proxy series and reaches up to the decade 1990–1999. The golden line is thermometer measurements from 1850; in the period they share, the two agree well.
What you see: The Roman Warm Period (roughly year 1–300) and the Medieval Warm Period (roughly 800–1300) are at the level of the 1961–1990 mean or slightly above it. In between lies a cold phase, followed by the Little Ice Age (roughly 1300–1900) with its low point around 1700. In these two millennia the temperature rose and fell quickly several times.
💬 In plain language
- The question is: is the temperature rising faster today than before?
- The answer depends on how long the period is. Over 50 years, a similarly strong rise has happened before (around 1700 and from 1950 to 1999). Over the last 30 years, the thermometer data show the highest rates.
- But: old data are smoothed and show short rises only weakly. So the two cannot be compared exactly.
- The ice core is too coarse for short periods. It only shows changes over millennia.
Is the temperature rising faster today than before?
The answer depends on the period over which the rate is calculated and how finely the data are resolved in time. Here are all the rates that can be calculated from the data on this page, in °C per century (extrapolated from the respective period).
| Data series | Period | Length | Rise per 100 years | Resolution |
|---|---|---|---|---|
| Vostok ice core (Antarctica) | about 11,000–12,000 years ago | 1,000 years | +0,34 | 500 years |
| Vostok ice core (Antarctica) | Transition about 130,000–135,000 years ago | 5,000 years | +0,17 | 500 years |
| Northern Hemisphere reconstruction (Ljungqvist) | 1690–1739, strongest rise before 1850 | 50 years | +0,95 | decade, proxies |
| Northern Hemisphere reconstruction (Ljungqvist) | 1700–1729, strongest rise before 1850 | 30 years | +1,41 | decade, proxies |
| Northern Hemisphere thermometers (Ljungqvist file) | 1950–1999 | 50 years | +0,91 | decade, measured |
| Northern Hemisphere thermometers (Ljungqvist file) | 1970–1999 | 30 years | +2,41 | decade, measured |
Blue: reconstructed from proxy data (ice, tree rings, sediments), gold: measured (thermometers). Calculated as a linear slope over the respective period.
What follows from this – expand explanation
What follows from this
- Over 50 years this has happened before: In the reconstruction, the strongest 50-year rise before 1850 (1690–1739, recovery from the Little Ice Age) was +0.95 °C per century, as large as the thermometer rise of 1950–1999 (+0.91).
- Over 30 years the measured rate is higher: The thermometer values for the last 30 years of the file (+2.41) are above the strongest 30-year rise in the reconstruction (+1.41). However, the two are not directly comparable: the reconstruction is assembled from proxies and smoothed over decades, the thermometer series is not. Smoothed curves show short, fast rises more weakly.
- The ice core cannot resolve decades: It only shows changes over millennia (a few tenths of a degree per century, in Antarctica). It says nothing about short rises.
In short: For periods of 50 years and longer, comparably strong rises have happened before. For short periods of about 30 years, the thermometer data today show the highest rates in the series available here, but the older data are too coarsely resolved to say whether similarly fast short rises occurred earlier. These data show neither that it has never gone this fast, nor that it was always like this.
Data as of November 2026. The data series end in different years depending on the source (mostly 2024 or 2025); the end point is given in the source note under each chart, an overview is on “Data and methods”.
