Assessment of CMIP5 and post-CMIP5 Representation of Tropical Cyclones in Forced Atmosphere Simulations - Understanding Dynamical Constraints

Explicit simulation of tropical cyclone properties, such as interannual variability and track, is a challenge for global climate models such as those in the Coupled Model Intercomparison Project (CMIP5) class, due to constraints on model resolution and other aspects. However, given the large number of existing simulations, together with new ones being prepared for CMIP6, it may be possible to understand some common model biases, and perhaps their relation to model resolution or other factors. This study uses output from the CMIP5 AMIPII present day simulations, together with both additional models post-CMIP5 and reanalyses, to study explicitly tracked tropical cyclones, using a common algorithm. This provides a range of horizontal resolutions between about 200km in several CMIP5 models, up to 14km in the Japanese NICAM model.

Current climate models vary broadly in terms of their skill in representing tropical cyclone frequency, intensity (as measured by mean sea level pressure), genesis and track position. The North Atlantic region is where many models have a reasonable level of correlation of interannual frequency with observations, with much lower skill in other basins. However, a common bias in many of models is the lack of storm genesis in the Eastern Atlantic, a region where the strongest U.S. landfalling storms are often observed to form and hence an important region to represent robustly in climate projections.

In an attempt to understand the impact that dynamical factors have on this bias, we have assessed the simulation of the African Easterly Jet (AEJ) and associated Waves (AEWs) in all the models, again using a common algorithm that allows the waves to be counted and their properties compared. We find evidence that the latitude of the AEJ has an influence on the curvature vorticity of the AEWs, and that in turn the frequency of tropical cyclones in the Eastern Atlantic is related to this curvature vorticity.

Given the uncertainty in future projections of tropical cyclone activity using different methods, it is important to better isolate and understand what constrains storm frequency in present climate conditions. To this end, the HighResMIP protocol for the Coupled Model Intercomparison Project (CMIP6) plans to study processes such as these in more detail, using both a longer forced atmosphere historic period (1950-2014) than has been typical previously, together with both standard and high (~25km) model horizontal resolutions, but with minimal model changes. It is hoped that such a constrained multi-model ensemble will enable both improved representation of tropical cyclones, and a database to improve our understanding of the driving factors.