Last updated July 5, 2018 at 9:47 am
Researchers warn that climate models may not be right.

Credit: Ben Goode / iStock
If the past is anything to go by, expected future global warming will see climate zones spatially shift and, on millennial time scales, ice sheets substantially shrink.
And that means many climate models designed to project changes within this century may underestimate longer-term changes, according to an international team which studied selected intervals in history that were as warm or warmer than today.
Their sombre finding is that even with global warming limited to within 2°C above pre-industrial levels, as envisaged in the Paris Agreement, climate zones and ecosystems will shift, rapid polar warming may release additional greenhouse gases, and sea level will rise by several metres over several thousand years.
The study, which involved 59 scientists from 17 countries, was coordinated by the University of Bern, the University of New South Wales and Oregon State University.
Stronger regional warming at high latitudes
Several time periods over the past 3.5 million years are known for being 0.5-2°C warmer than the so-called pre-industrial temperatures of the 19th century. These intervals reveal much stronger regional warming at high latitudes than in the tropics, similar to what models predict for a 2°C global warming by the year 2100.
Although not all past warmings were caused by higher CO2 concentrations, the researchers say they are helpful in assessing the regional effect of a warming of a scale comparable to that aimed at in the Paris Agreement.
According to Associate Professor Katrin Meissner from the University of NSW, comparison of observations of the past with computer simulations suggests that models may underestimate long-term warming and its amplification in polar regions.
“While climate model projections seem to be trustworthy when considering relatively small changes over the next decades, it is worrisome that these models likely underestimate climate change under higher emission scenarios, such as a ‘business as usual’ scenario, and especially over longer time scales,” she said
The paper published in Nature Geoscience.