Melissa Moves Away, West Pacific Heating Up: AI Models Discussion 10-31-2025

Atlantic

I am returning back to a normal routine after about a week of flying with the NOAA Hurricane Hunters to collect data in Hurricane Melissa. This was a historical set of flights, including one mission Tuesday morning right before landfall that found a strong Category 5 storm that tied the record for lowest pressure at landfall in the Atlantic. It was very sobering to see such a powerful storm on the doorstep of the island of Jamaica, and I hope the recovery efforts there will be swift and effective. Today, Melissa is winding down over the North Atlantic and will soon be an extratropical cyclone, though it may bring some rain and wind to Newfoundland and the Canadian Maritimes, after already impacting Jamaica, Haiti, the Dominican Republic, Cuba, Bahamas, Turks and Caicos, and Bermuda. Sometime during the next week or so I will plan to do a blog exploring the forecast performance of the various AI models for Melissa.

In the wake of Melissa, the Atlantic is quiet for now, with models not showing any signs of development for the next week or so. With cold fronts starting to come down into the Gulf and even Caribbean, we should start to see the seasonal decrease in SST and increase in shear that marks the start of the end of hurricane season. However, it’s always possible one of these fronts could enhance low-level convergence and try to spark development in the Caribbean, but there are no indications of that at the moment.

West Pacific

In contrast to the Atlantic (which is quieting down), the West Pacific seems to be heating up, as an MJO pulse moves into the region. Ensembles suggest that two potentially strong systems could form over the region during the next 7-10 days (Figure 1).

The first system, Invest 98W, is located to the East of the Southern Philippines. Some models and ensemble members show this system intensifying quite a bit before reaching the Philippines. For example, FourCastNet-V3 shows a solid hurricane in 3 days just east of the Philippines (Figure 2).

Others are slower to develop prior to reaching the Philippines, such as GraphCast (Figure 3). Some of the intensification may depend on how long it is able to stay over water as it works its way near and through the islands of the Philippines. It will be worth keeping a close eye on this system to see how quickly it is able to get its act together over the next day or two as it moves west.

After crossing the Philippines, almost all models/ensembles take it across the South China Sea towards Vietnam, intensifying in the process. Upper-level winds look to be generally favorable with a large anticyclone overhead and to the north (Figure 4). Quite a few Deepmind/FNV3 members make this a strong typhoon approaching Vietnam, so this system is definitely one to watch there next week.

The second system (Invest 99W) is located further east, and also forecast to move generally toward the west over the next several days. There is a strong signal in Deepmind and other guidance for this to become a potentially powerful typhoon in the warm water and favorable atmospheric conditions east of the West Pacific warm pool east of the Philippines. One of the big forecast questions in the medium range is how far west the potential cyclone will get. Some guidance suggests that the system will be trapped under a ridge and approach Luzon (Figure 5), while other models indicate that the ridge may weaken as a trough passes to the north (Figure 6). Ensembles suggest both possibilities are in play, and there is a lot of uncertainty given this is a 7-10 day forecast. It will definitely be worth keeping an eye on this system heading into next week, to see how quickly it develops, how the synoptics evolve, and which land areas it will threaten, if any.