Attributing and Projecting Heatwaves Is Hard: We Can Do Better
Heatwaves are arguably the most deadly of weather phenomena. As the earth warms due to higher concentrations of greenhouse gases, one would expect heatwaves to become worse as well, killing even more people unless they are better protected against the heat. However, it turns out that the world is not so simple and that many other factors also influence heatwaves. Land-use changes, irrigation, air pollution, and other changes also drive trends in heatwaves. Some of these cause much larger trends while some have counteracted the climate change-driven trends up to now. In some regions, the causes of high trends have not yet been identified. Current generation climate models often do not simulate all these mechanisms correctly so will have to be improved before we can more confidently trust their description of past trends and projections of future trends in heatwaves.
There remains an uncomfortably large list of potential reasons for our current lack of understanding of the drivers of extreme heat, including land-use changes and soil moisture, aerosol effects, and atmosphere feedbacks as well as circulation effects other than blocking. There remain three possible reasons for divergence between observed and simulated heat extremes at the scales that affect people. First, the possibility that the models are right, but are being given incomplete local information such as missing land surface feedbacks and use changes, etc. Second, the possibility that the models are truly incorrect and would not have captured observed trends even if these regionally-specific matters were fully incorporated into the modeling framework. Third, that natural variability at local scales predominates over anthropogenic forcing and that the models either do not simulate internal variability correctly or our ensembles are not large enough to capture it.
While large-scale changes in mean temperature are well understood, changes in local and regional heatwaves, particularly, daytime maxima, are much harder to simulate and hence attribute. This failure to understand today's observed trends and the discrepancies between the modeled and observed trends and variability also hinders confidence in projections of the future trends. The extrapolation of the observed trends often is very different from the simulated trends from climate model output over the same period. In our view, it is thus an immensely important priority for climate model development studies to focus on extreme heat, the deadliest and most immediate effect of human-induced climate change.