Increasing the concentration of carbon dioxide (CO2) in the atmosphere traps more heat energy, causing global surface temperatures to rise. This is the direct effect of rising CO2 levels, which scientists can quantify with laboratory measurements and basic physical calculations. But there are also indirect effects caused by climate feedbacks, which either increase or reduce an initial change in temperature. Climate model studies and atmospheric measurements indicate that the overall effect of climate feedbacks is to increase a temperature change. However, the size of this amplification is more difficult to quantify. Examining how global temperatures responded to past changes in the energy balance helps scientists predict how the climate could respond to the present CO2-induced energy changes.

Any change in global average surface temperature is triggered by a change in the Earth’s energy balance. This is caused either by a change in the amount of solar energy being absorbed by the Earth’s surface or by a change in the strength of the greenhouse effect, which dictates how much of the Earth’s heat energy escapes to space. More energy entering the climate system than leaving it results in an energy surplus, causing temperatures to rise. More energy leaving the climate system than entering it results in an energy deficit, causing temperatures to fall. Past changes in the Earth’s energy balance have been caused by shifts in the Earth’s orbit round the Sun, variations in solar output and changes in the concentrations of greenhouse gases in the Earth’s atmosphere.

Scientists use data from climate proxies to investigate past climate changes. By estimating the magnitude of the initial change in the Earth’s energy balance that triggered a previous episode of warming or cooling, along with the magnitude of the temperature change that followed, scientists can compare the two. There are many uncertainties in this process, but if both quantities can be estimated with sufficient confidence, scientists can estimate how much of the temperature change was caused by the direct effect of the initial change in energy, and how much was caused by feedback effects. This can help scientists estimate the overall strength of temperature feedbacks, which in turn can help to estimate the climate’s response to a doubling of atmospheric carbon dioxide – the ‘climate sensitivity’.

Some scientists have estimated ‘climate sensitivity’ by looking at the energy imbalance and consequent temperature rise in the 20th century. Such estimates tend to cluster around a sensitivity value of 2–3 °C, similar to the ‘best estimate’ values derived from other methods. But these calculations have large uncertainties because of the difficulties of estimating the overall energy imbalance. This is because mid-century human activities emitted a lot of aerosols which reflect sunlight, offsetting some of the energy surplus from increasing greenhouse gases. Estimating climate sensitivity from 20th-century temperature change is also complicated by the climate system’s time-lag, which means that temperatures are still adjusting to the energy imbalance caused by rising greenhouse gases.

Using data from proxies, scientists calculate that global temperatures during the coldest part of the last ice age were between 3 and 9 °C colder than today, with a most likely difference of about 5–6 °C. Triggered by shifts in the Earth’s orbit, the ice age owed its low temperature to a higher albedo and a weaker greenhouse effect resulting from climate feedbacks. Analysis of air bubbles trapped in ice cores lets scientists measure ice-age greenhouse gas levels. Assessments of the likely differences in ice cover, dust and vegetation at the time lets scientists estimate the Earth’s ice-age albedo. Scientists then calculate the resulting energy imbalance. Comparing this with the energy imbalance for a doubling of carbon dioxide implies that ‘climate sensitivity’ lies in the range 1.3–4.6 °C, with a ‘best estimate’ of 2.7 °C.

There are substantial uncertainties involved in estimating past temperature changes and the changes in energy that triggered them. This leads to a wide uncertainty range when using past changes to estimate ‘climate sensitivity’. Sensitivity values calculated using this method range from as low as 1 °C to as high as 10 °C, though most scientists consider both these extremes to be implausible. The ‘best estimate’ range of climate sensitivity derived from past climate changes is about 2–4.5 °C – in broad agreement with the results of climate models and the estimates made by using basic physics. Climate processes may not behave the same way in the future as in the past, so estimates of sensitivity from past climate change can only be used as a guide, for comparison with other prediction methods such as climate modelling.

Jane Francis is a palaeoclimatologist who studies past climates. Her research shows that the Earth was warm enough for forests to grow in polar regions millions of years ago. ‘This research can help us understand how our climate might change in future as our world gets warmer.’ Spending her time at Leeds University and out in the field in Antarctica, Jane collects fossils to understand what vegetation was growing millions of years ago. ‘Using these fossils I’ve been able to reconstruct what polar forests looked like millions of years ago – a bit like a detective, piecing the story together from bits of evidence.’