"Over the current decade (specifically, for eight years from 2000 to 2007), the globally-averaged intensity of heat waves calculated from observations is higher and shows a more increasing trend compared to even the worst case projections from climate models," said Auroop Ganguly of Oak Ridge National Laboratory, US. "Increased trends point to the urgency of international and domestic policy negotiations for reducing emissions."

Ganguly and colleagues at Oak Ridge, the University of Notre Dame, and the US National Center for Atmospheric Research used ensemble simulations from a global earth systems model to analyse bias and uncertainty. They compared this data with observations for 2000–2007.

"The larger uncertainties, both at global and regional scales, suggest that policymakers and stakeholders may face a more complex task to decide the various societal cost-benefit trade-offs during decisions on climate related mitigation and adaptation related spending," said Ganguly. "Uncertainty is not an excuse for inaction, but an opportunity for more cautious and risk-informed decisions."

On the one hand, says Ganguly, increased uncertainty may imply a worst-case where temperature extremes are much more severe and frequent that even the most dire projections. "This implies a need to plan accordingly for a much higher worst-case, for example where human lives may be involved (e.g. consider the human mortality from the Paris and Chicago heat waves)," he said. "On the other hand, the larger uncertainty may also imply that the temperature extremes may not be as severe or frequent as we thought from the mean projections."

Such considerations may help in deciding cost-benefit trade-offs in certain situations, for example, the uncertainties in a shift in water resources or agricultural patterns in the Western US may need to be taken into account during cost-benefit analysis of climate change risk-management approaches.

The team also found that although there is often an increase in the severity of heat waves when a region experiences warming, the occurrence or severity of heat waves does not necessarily correspond to the regions where the highest regional warming occurs. "This is counter-intuitive, but extreme events do not necessarily co-occur with large regional changes in climate," said Ganguly.

The researchers began their study following participation in a climate change war game in 2008, for which they prepared maps of temperature increases in the 21st century. "The lessons learned from the war game motivated us to look deeply into the uncertainty associated with the projections of climate change and extremes, especially at regional scales in space and decadal scales in time," said Ganguly.

Now the researchers plan to address the limitations in this study; analyse trends in extremes, regional change and uncertainties for variables such as precipitation and wind; generate predictive insights for the statistics of extremes such as tropical cyclones, wildfires, landslides, sea-level rise, sea ice, and land glacier melts, which can be inferred with varying uncertainties from ancillary climate models; and analyse the consequences of extreme hydrometeorological events or large regional shift in climate patterns using information about vulnerability from projections of population or economic growth.

The researchers reported their work in PNAS.