Climate models tend to overestimate the extent to which climate change contributes to weather events such as extreme heat and rain.

Omar Bellprat and Francisco Doblas-Reyes at the Catalan Institute of Climate Sciences in Barcelona, Spain, used an idealized statistical model to compare the frequency of weather extremes in simulations with and without climate warming. Extreme events seemed to be more closely linked to climate change when the model was forced to run at low levels of reliability than when the model error was kept to a minimum.

To account for models' biased representation of climate variability, studies should rely on calibrated model ensembles, which are commonly used by weather forecasters, the authors suggest.

Geophys. Res. Lett. http://doi.org/bcsr (2016)