"The upper end of the climate-model temperature projections appears to be inconsistent with past warming," Peter Stott of the Met Office Hadley Centre, UK, told environmentalresearchweb. "This means that those models that warm fastest appear to be warming too rapidly, according to recent observations."

Stott said the CMIP5 model range appears to be correct in projecting substantial warming over the coming century with continued anthropogenic emissions of greenhouse gases. "This contradicts claims from some quarters that the recent hiatus in global mean warming indicates that much less warming is expected in future than was previously thought," he added.

The study by Stott and colleagues from the Met Office Hadley Centre, Environment Canada and the University of Reading, UK, indicated that the 5–95% range of warming by the 2020s relative to 1986–2005 under the RCP4.5 scenario should be 0.35–0.82 K, rather than 0.48–1.00 K. Similarly, under the RCP8.5 scenario, the range should be 0.45–0.93 K rather than 0.51–1.16 K. The RCP (Representative Concentration Pathway) 4.5 scenario, which assumes that radiative forcing in the year 2100 will be 4.5 W/sq m, will be used in the IPCC's fifth assessment report.

"Care needs to be taken in interpreting the results of ensembles of climate models, which are ensembles of opportunity rather than carefully designed experiments to sample modelling uncertainty," said Stott.

The researchers said they employed a standard technique that has been used before, based on detection and attribution of climate change, but updated previous studies by looking at data from the first decade of the 21st century. Improved estimates of likely future warming rates could result from better representation of internal variability and the response of the climate system to external forcings, as well as additional model simulations including future increases in well-mixed greenhouse gases.

Simply waiting for another 10 years of near-surface temperature data should reduce the uncertainties, wrote the team in Environmental Research Letters (ERL) , but use of other observations too, for example temperatures of the interior of the ocean, may help to improve such constraints sooner and identify which aspects of climate-model simulations of the past may be in error, be it ocean-heat uptake, climate sensitivity or net forcing.

"It would be interesting to see if it is possible to provide improved estimates of likely future rates of warming in different regions of the world," said Stott, "as well as seeing to what extent it might be possible to use observations to better understand future changes in other climate variables such as sea-ice reduction and rainfall changes."

Related links

Related stories