Weather has long been projected to possess limited predictability due to the inherent chaotic nature of the atmosphere; small changes in initial conditions could lead to an entirely different state of the atmosphere after some period of time. Given such a limited range of predictability of atmospheric flows, a natural question is, how far in advance can we predict a hurricane’s intensity? In this study, it is shown first that the predictability of a hurricane’s intensity at the 4–5-day lead times is generally determined more by the large-scale environment than by a hurricane’s initial conditions. This result suggests that future improvement in hurricane longer-range intensity forecasts by numerical models will be most realized as a result of improvement in the large-scale environment rather than in the storm’s initial state. At the mature stage of a hurricane, direct estimation of the leading Lyapunov exponent using an axisymmetric model reveals, nevertheless, the existence of a chaotic attractor in the phase space of the hurricane scales. This finding of a chaotic maximum potential intensity (MPI) attractor provides direct information about the saturation of a hurricane’s intensity errors around 8 m s−1, which prevents the absolute intensity errors at the mature stage from being reduced below this threshold. The implication of such intensity error saturation to the limited range of hurricane intensity forecasts will be also discussed.

This study presents a minimal model to examine the stability of the hurricane maximum potential intensity (MPI) equilibrium, based on the evolution of several hurricane fundamental scales. It is shown that this hurricane-scale model contains MPI as an asymptotically stable spiral point under the wind-induced surface heat exchange (WISHE) feedback mechanism. Furthermore, the hurricane development is characterized by two different time-scales instead of a simple linear growth, as often assumed in previous studies. Inclusion of radiative cooling reduces the MPI limit, but the MPI equilibrium in general is structurally stable. In particular, radiative cooling will modify the basin of attraction of the MPI equilibrium and create a domain in the phase space of the hurricane scales such that a vortex would not develop if its initial condition falls within this domain. Potential mechanisms that can further unfold the stability of the MPI equilibrium will also be discussed.

This study examines the large‐scale factors that govern global tropical cyclone (TC) formation and an upper bound on the annual number of TCs. Using idealized simulations for an aquaplanet tropical channel, it is shown that the tropical atmosphere has a maximum capacity in generating TCs, even under ideal environmental conditions. Regardless of how favorable the tropical environment is, the total number of TCs generated in the tropical channel possesses a consistent cap across experiments. Analyses of daily TC genesis events reveal further that global TC formation is intermittent throughout the year in a series of episodes at a roughly 2‐week frequency, with a cap of 8–10 genesis events per day. Examination of different large‐scale environmental factors shows that 600‐hPa moisture content, 850‐hPa absolute vorticity, and vertical wind shear are the most critical factors for this global episodic TC formation. Specifically, both the 850‐hPa absolute vorticity and the 600‐hPa moisture are relatively higher at the onset of TC formation episodes. Once TCs form and move to poleward, the total moisture content and the absolute vorticity in the main genesis region subside, thus reducing large‐scale instability and producing an unfavorable environment for TCs to form. It takes ∼2 weeks for the tropical atmosphere to remoisten and rebuild the large‐scale instability associated with the Intertropical Convergence Zone before a new TC formation episode can occur. These results offer new insight into the processes that control the upper bound on the global number of TCs in the range of 80–100 annually.

This study examines the formation of a double warm-core (DWC) structure in intense tropical cyclones (TCs) that was captured in almost all supertyphoon cases during the 2012–14 real-time typhoon forecasts in the northwestern Pacific basin with the Hurricane Weather Research and Forecasting Model (HWRF). By using an idealized configuration of HWRF to focus on the intrinsic mechanism of the DWC formation, it is shown that the development of DWC in intense TCs is accompanied by a thin inflow layer above the typical upper outflow layer. The development of this thin inflow layer in the lower stratosphere (~100–75 hPa), which is associated with an inward pressure gradient force induced by cooling at the cloud top, signifies intricate interaction of TCs with the lower stratosphere as TCs become sufficiently intense, which has not been examined previously. Specifically, it is demonstrated that a higher-level inflow can advect potentially warm air from the lower stratosphere toward the inner-core region, thus forming an upper-level warm core that is separated from a midlevel one of tropospheric air. Such formation of the upper-level warm anomaly in intense TCs is linked to an episode of intensification at the later stage of TC development. While these results are produced by HWRF, the persistent DWC and UIL features in all HWRF simulations of intense TCs suggest that the lower stratosphere may have significant impacts on the inner-core structures of intense TCs beyond the current framework of TCs with a single warm core.

Exploring new techniques to improve the prediction of tropical cyclone (TC) formation is essential for operational practice. Using convolutional neural networks, this study shows that deep learning can provide a promising capability for predicting TC formation from a given set of large-scale environments at certain forecast lead times. Specifically, two common deep-learning architectures including the residual net (ResNet) and UNet are used to examine TC formation in the Pacific Ocean. With a set of large-scale environments extracted from the NCEP–NCAR reanalysis during 2008–21 as input and the TC labels obtained from the best track data, we show that both ResNet and UNet reach their maximum forecast skill at the 12–18-h forecast lead time. Moreover, both architectures perform best when using a large domain covering most of the Pacific Ocean for input data, as compared to a smaller subdomain in the western Pacific. Given its ability to provide additional information about TC formation location, UNet performs generally worse than ResNet across the accuracy metrics. The deep learning approach in this study presents an alternative way to predict TC formation beyond the traditional vortex-tracking methods in the current numerical weather prediction. Significance Statement: This study presents a new approach for predicting tropical cyclone (TC) formation based on deep learning (DL). Using two common DL architectures in visualization research and a set of large-scale environments in the Pacific Ocean extracted from the reanalysis data, we show that DL has an optimal capability of predicting TC formation at the 12–18-h lead time. Examining the DL performance for different domain sizes shows that the use of a large domain size for input data can help capture some far-field information needed for predicting TCG. The DL approach in this study demonstrates an alternative way to predict or detect TC formation beyond the traditional vortex-tracking methods used in the current numerical weather prediction.