Process-based pathways to improving projections of extremes

Extreme precipitation, droughts and heat waves are among the most impactful effects of climate change— improving the precision of climate model projections of changes in these events is a key endeavor. While higher resolution models may well help, here we focus on examples where understanding of climate process mechanisms can narrow uncertainty. In some situations this can be done for post-simulation analysis, in others it can provide input for improving process representation in climate models, point out physical pathways for attention, or provide fundamental underpinnings for understanding emergent effects. Examples are provided to overview several effects. The “dry-inhibition regime” in which dry air in the lower free troposphere (LFT) inhibits development of precipitating deep convection even when the surface is experiencing warm and moist extremes can affect both moist heat waves and dry intervals. It depends on entrainment processes— both turbulent and at the scale of the convective entity — that incorporate LFT air. This has been a long standing challenge for climate models to represent, but a combination of process modeling and observations is narrowing this gap. Understanding of how precipitation probability distributions arise at a process level indicates that climate models do better at certain aspects of extreme precipitation probability distributions than previously thought although substantial uncertainties remain in the regional scaling of probability increases of intense precipitation. This may be associated with long-standing challenges in understanding the effective moist stability of the atmosphere in precipitating conditions. Finally, theory for autocorrelated temperature evolution provides a basis for climate model analysis showing that probabilities of long heatwave durations have an unavoidable acceleration per increment of warming. The same analysis allows collapse of climate model future projections across regions onto scaling that can be compared with reanalysis warming in the historical period, reinforcing confidence in this effect.