Large-scale Influences on Tropical Cyclone Activities in a Superparameterized GCM Aquaplanet Experiment

Understanding and predicting tropical cyclone (TC) activities are important for mitigating TC risks to human society. However, general circulation models (GCM) are integrated at moderate horizontal resolutions (~50 km) in climate projections; TC characteristics in these runs tend to be unrealistic and very sensitive to how convection is parameterized. The so-called multiscale modeling framework provides a way to improve GCMs by inserting a cloud resolving module- but only within each model grid column to “super-parameterize” convective-related processes.

Here we compared the TC activities in the super-parameterized Community Atmosphere Model (SPCAM) v5 with those in the convectively parameterized CAM (CPCAM, v5; Zhang and McFarlane deep convection and Park and Bretherton shallow convection schemes), in aquaplanet, perpetual summer experiments. It was found that SPCAM had a TC frequency 4-5 times higher than that produced by CPCAM. In SPCAM the genesis potential index and 500-hPa pressure velocity were also significantly stronger, while the SPCAM ventilation index was found to be weaker off the equator than that in CPCAM, indicating an environment more conducive for TC genesis. More TC “seeds” were also found in SPCAM compared to CPCAM. For individual TCs, there were stronger maximum 10-m wind speed and lower minimum sea level pressure, while the wind-pressure relationship was also more realistically captured, in SPCAM compared to CPCAM. Further inspection of the GCMs’ general circulation revealed an Inter-topical Convergence Zone (ITCZ) farther (closer) to the equator, with more (less) frequent ITCZ breaking in SPCAM (CPCAM). SPCAM also exhibited enhanced equatorial Rossby and mixed Rossby-gravity wave activities, as well as tropical depression-type systems, in its model environment. Implications were drawn for applying a multiscale modeling framework in GCMs to reduce uncertainties in climate and extreme event modeling.