The team, led by Dr Charles Williams, Senior Lecturer in the School of Geographical Sciences, wanted to investigate factors influencing the way the Earth’s climate has repeatedly swung between cold glacial ‘ice ages’ and warmer interglacial periods over the last 2.6 million years – known as the Quaternary period.
Scientists have long known that these cycles are linked to subtle changes in Earth’s orbit around the Sun, combined with feedback processes within the climate system, such as changes in atmospheric carbon dioxide (CO₂) and the growth and retreat of massive ice sheets.
However, recreating these climate cycles using traditional global climate models is extremely demanding. Running such models over millions of years requires vast computing resources and can take decades of real time.
In a new study published in Nature Communications, researchers overcame this challenge by using a powerful climate model to train a much faster statistical emulator that can reproduce the behaviour of the full model. Once trained, the emulator can simulate climate changes across the entire Quaternary period in just minutes and run on a standard laptop.
“Even a relatively fast climate model would take around 60 years of real time to simulate three million years of climate,” explained Dr Williams. “Our emulator can perform the same simulation in about 10 minutes.”
The team first tested the emulator by comparing its results with geological proxy records covering the past 800,000 years. These records, derived from sources such as ice cores and ocean sediments, provide evidence of past temperatures and ice volumes. The emulator successfully reproduced the timing and scale of major ice-age cycles seen in the geological data.
With confidence in the model’s performance, the researchers then ran a series of experiments to examine what factors drive long-term climate change.
Their findings support existing scientific understanding: while variations in Earth’s orbit help set the timing of ice ages, internal feedbacks within the climate system play the dominant role in determining how large these climate swings become. In particular, interactions involving atmospheric CO₂ and the expansion and contraction of ice sheets were found to be the primary drivers of global temperature changes over million-year timescales.
“What makes this work exciting is the methodology,” Dr Williams said. “Statistical emulators have been used in palaeoclimate studies before, but to our knowledge they’ve never been applied to simulate the entire Quaternary period.”
The new approach opens the door to studying climate processes on timescales that were previously impractical to simulate. Because the emulator runs so quickly, scientists can also perform many more experiments, such as switching individual climate drivers on and off, to isolate their influence on global climate.
This could help researchers better understand the complex interactions that shape Earth’s long-term climate system, while dramatically reducing the cost and computing power typically required for such studies.
“This technique could transform how scientists investigate climate change over hundreds of thousands to millions of years, making it possible to explore long-term climate dynamics in far greater detail than before,” Dr Williams concluded.
Paper:
‘The relative role of direct orbital forcing versus CO2 and ice feedbacks on Quaternary climate’, by C.J.R. Williams et al in Nature Communications