Lagrangian Analysis of Tropical Cyclone Intensification Simulated by General Circulation Models Compared with Observations

Abstract

Abstract To understand the intensification process of tropical cyclones (TCs), we analyzed the relationship between the TC intensification rate (I) and environmental variables along TC tracks during the time from TC genesis (tG) to maximum TC strength (tX), hereinafter τGX ≡ tX − tG, using a state-of-the-art general circulation model (GCM), observed TC tracks, and ERA5 data. During τGX, strong TCs with high I (sTCs) consume more convective available potential energy (CAPE) than weak TCs with low I (wTCs) and bring more CAPE from the equator to sustain sTCs. Compared to wTCs, sTCs prefer an unstable atmosphere with higher sea surface temperature (SST), stronger grid-mean upward flow at 500 hPa (ω500), more moisture convergence (MC), and weaker wind shear (Vs). Our GCM simulation shows that MC and CAPE have a single regression slope with I applicable both within and across climate regimes. Using machine learning, we found that the best combination of environmental variables (V6) for predicting I consists of ω500, MC, SST, midtropospheric stability (MTS), Vs, and latitude (|f|). Machine learning with V6 reproduces well the spatial distribution and interclimate changes of I: TCs are intensified in regions of stronger upward ω500, more MC, warmer SST, weaker MTS, smaller Vs, and larger |f|; TCs in a warmer climate have higher I than TCs in a colder climate due to more MC, warmer SST, but stronger MTS. These results are consistent with the conceptual understanding that TCs are intensified by the release of latent heat.