Next let's compare it to a warm seclusion that later did successfully undergo tropical transition - a rare case in January 2023. Not only did it fully detach from frontal boundaries & baroclinic forcing for a much longer period of time, but it also sustained its own convection.

Now that the nor'easter completed its phase with a tropopause vortex and is decidedly non-tropical, let's look back at its structural evolution over the last day and a half, and why despite attempting to start the tropical transition process, it fell short of classification:

Go home GFS you're drunk again

A few high-resolution convection allowing models come close - the experimental GSL MPAS as shown below, for example, does seem to try to undergo tropical transition - but this is a minority solution ATTM & would require a deeper cyclone quickly secluding while over warmer SSTs.

Phase space diagrams from most models - like the ECMWF - identify a shallow warm core, which by itself is insufficient to classify a tropical cyclone. The cyclone would need to detach from frontal boundaries while maintaining a moderate-deep warm core & persistent convection.

Having written this kind of algorithm myself, I can speak from experience that it's challenging writing an objective method using only coarse global model fields to estimate what is or isn't a tropical cyclone in ensemble members, then validating it across many false alarms.

This is the time step when ECMWF's objective algorithm has AIFS peaking in intensity while cateogorized as a tropical cyclone. A simple look at this map is a clear tell to any meteorologist or experienced forecaster that this is not a tropical cyclone:

To illustrate this, let's look at the 12z AIFS - which ECMWF's tracker identified as a tropical cyclone at hour 48. Looking at its model fields, this is clearly a baroclinic cyclone associated with synoptic forcing undergoing warm seclusion:

All of the plots in the post above this are from objective algorithms that use global model & ensemble member data to estimate whether a cyclone is tropical or non-tropical. This is no easy task, and can mis-identify systems often. Per ECMWF's own explanation:

Google DeepMind shows a 33% chance of a tropical storm. ECMWF's tropical cyclone probabilities from EPS & AIFS exceed 80%. Surely that means this nor'easter will be a tropical storm, right? Short answer... it's not that simple. Breaking down why in this technical thread 🧵

The fact this was a tropical depression just 48 hours ago is absolutely mind blowing

A less well-known fact about this hurricane is that it hit on Yom Kippur and on a Saturday, as Miami's still-nascent Jewish community was growing. Many left Yom Kippur services as conditions were worsening, and some persevered fasting through destructive conditions.

AI’s ceiling may be far beyond what most people can comprehend. That doesn’t mean “most people are stupid”. It means “it is genuinely difficult to envision an end point to something which has no precedent”. Think of it like asking someone 150 years ago what technology in 2026 will look like.

Initial cyclogenesis is largely forced downstream of the upper-level trough as it cuts off. Whether this undergoes tropical transition & to what extent depends partly on where & how quickly the initial cyclone develops - 0z ECM (left) forms a medicane while the 6z (right) doesn’t.

The key driver of this Medicane potential is anticyclonic wavebreaking over Europe leading to a cutoff low over the southern Mediterranean — right near unusually warm waters over 30C, which is more than sufficiently warm to support a tropical cyclone.

There is finally a potential for a hurricane or strong tropical cyclone east of North America — but it’s in the Mediterranean. There’s been a growing signal among ensembles for a medicane, or subtropical/tropical cyclone in the Mediterranean, that may develop this week.

I also added a few more updates by request, such as customizable forecast labels and a distance measurement tool: