Extreme temperature and precipitation events are expected to increase in frequency under climate change. However, projections of these events at the local to regional scale have large uncertainties. A study published in Nature Climate Change this week reports that these uncertainties are mainly due to internal variability of the climate system, and will remain even with model improvements. Despite this, the work reports that averaging across regions allows robust projection of future extremes.

Erich Fischer, Reto Knutti and colleagues use 25 models from the Coupled Model Intercomparison Project Phase 5 to simulate temperature and precipitation extremes for the periods 2016-2035 and 2041-2060. They find that all extremes have large uncertainties, however those associated with precipitation are the largest. To determine the role of internal variability, multiple runs of a single Earth system model were performed with only the initial climate conditions altered. Internal variability is shown to be responsible for 40-60% of uncertainty for temperature extremes, and over 75% for precipitation extremes. These results highlight that reliable information on changes in extremes at the local scale is not obtainable, but the authors note that the use of spatial averaging provides robust evidence of extremes in the next two to three decades. This information will be relevant for planning and adaption in the face of future climate events.