Regions which have, so far, not experienced a particularly extreme event may be less prepared for the consequences of such an event. We have identified the regions where the current records have the lowest return periods . In these regions, a record-breaking event is not only more likely but also likely to have greater impacts due to a lack of preparedness. Countries tend to prepare to the level of the greatest event they have experienced within collective memory.
Our global assessment of reanalysis data shows that statistically implausible extremes have occurred in 31% of regions between 1959 and 2021, with no apparent spatial or temporal pattern. It appears that such extremes could occur anywhere and at any time. When using climate model data to investigate further, we find 18–26% of regions in the model have the same characteristics.
The vulnerability of a region to the impacts of heat is not only dependent on the statistical likelihood of a record-breaking event. Socio-economic factors will make a large difference to the preparedness, with developing countries less likely to have adequate heat plans in place. Countries with greater projected population growth may be able to cope with current conditions but may find their health services and energy supply overwhelmed if policymakers do not plan adequately.
Although changing dynamics could be a factor in the more recent events, we have shown extremes beyond the statistical fit, exceptional extreme events, occur throughout the reanalysis time period—so this does not fully explain the outliers. Further investigation into whether the underlying statistics of the distribution are shifting, resulting in past observations no longer being useful for assessing future, or even current, risk of extremes, is needed.
As we show, many heat extremes are beyond the statistical distribution in both observations and the model data; it is not possible to estimate the most extreme plausible events using EVT. Further research into the greatest plausible climatic extremes is essential to allow policymakers to plan for possible future events.