West Pacific Heating Up? AI Models Discussion 5-18-2026

  • Andy Hazelton
    Andy Hazelton

Share post

West Pacific Heating Up?

The Northern Hemisphere tropical cyclone season is starting to wind up, and it looks like the first basin to take advantage may be the West Pacific. This could be a very active Typhoon season thanks to the developing El Niño event. Figure 1 shows the Google DeepMind TCs forecast over the next two weeks, showing a fairly robust signal for a TC to form between Guam and the Philippines.

Figure 1: Google DeepMind forecasts of TC tracks over the West Pacific, initialized at 12 UTC May 18, 2026, and valid out to 312h. Data from Google.

There is a lot of spread in track and intensity at this range, and TC formation is not guaranteed, though it makes some sense with the Madden Julian Oscillation (MJO) moving across the region. Other models also show some development, though it’s mostly in the Day 9-10 range right now (Figures 2-3).

Figure 2: AIGFS forecast of 850-hPa wind initialized at 12 UTC May 18, 2026, valid at 18 UTC May 27, 2026. Data from Tropical Tidbits.
Figure 3: AIFS forecast of mean sea level pressure and anomalies initialized at 00 UTC May 18, 2026, and valid at 06 UTC May 28, 2026.

The environment looks marginal for development. It looks like a tropical upper tropospheric trough (TUTT) will be draped across the West Pacific, but if it cuts off and a ridge builds overhead, the upper-level winds could become more favorable for development (Figure 4).

Figure 4: GraphCast forecast of 200-hPa wind initialized at 00 UTC May 18, 2026, valid at 12 UTC May 27, 2026.

On some models, however, the TUTT takes longer to cut off, and with more troughing over the West Pacific (Figure 5), TC development is delayed. This process of eroding the TUTT involves a lot of “diabatic heating” produced by convection, a process that can be difficult to model. So I will be curious to see how AI models handle this evolution.

Figure 5: FourCastNet-v3 forecast of 200-hPa winds initialized at 00 UTC May 18, 2025 and valid at 18 UTC May 26, 2026.

Regardless of development, it looks like this MJO passage will be associated with another period of westerly winds or reduced trades across the equatorial Pacific, helping to amplify the developing El Niño event (Figure 6).

Figure 6: GraphCast forecast of 10-meter wind initialized at 00 UTC May 18, 2026 and valid at 18 UTC May 22, 2026.

As the MJO pulse progresses eastward, we could see development in the East Pacific towards the start of June, and of course we’ll have to watch the Caribbean and Gulf for early-season development as well.