Bavi Moves West Towards East Asia, East Pacific Waking up Soon?

  • Andy Hazelton
    Andy Hazelton

Share post

Bavi Moves West Towards East Asia, East Pacific Waking up Soon?

With El Niño continuing to grow, the Pacific is where we find most of the current and forecast TC activity so far in July. We’ll go from West to East.

West Pacific

The West Pacific remains very active. Maysak moved inland in China, but Typhoon Bavi continues westward as a Super Typhoon with 140 knot (160 mph) winds, after making a direct hit on the island of Rota in the Mariana Islands yesterday. It is forecast to move west and then turn northwest towards Taiwan or Northeast China. There is also some signal for additional West Pacific development in its wake (Figure 1).

Figure 1: Google DeepMind 50-member ensemble forecasts of TC tracks and intensity over the West Pacific initialized at 06 UTC July 6, 2026, valid out to 360h. Data from Google.

One of the interesting forecast questions with Bavi is how close it will come to Taiwan. Google Deepmind’s ensemble is near-unanimous in keeping the TC north of Taiwan and continuing on towards the coast of China (Figure 2). Some of the dynamical models and ensembles (such as the ECMWF), however, bring the TC directly into the coast of Taiwan (Figure 3).

Figure 2: Google Deepmind ensemble forecasts for Typhoon Bavi initialized 06 UTC July 6, 2026, and valid out to 144h. Figure from www.weathernerds.org.
Figure 3: ECMWF ensemble forecasts for Typhoon Bavi initialized 00 UTC July 6, 2026, and valid out to 150. Figure from www.weathernerds.org.

The differences between the AI and dynamical models appear to start early, and are related to very subtle differences in steering. The ridging to the north of the TC is slightly stronger in the ECMWF (Figure 4) than the EC-AIFS (Figure 5), for example, as can be seen in the slightly larger area of 591-dam heights south of Japan. This difference leads to a more westward motion early in ECMWF, which allows the TC to impact or pass very close to Taiwan as it moves NW. On the AIFS, the slower and slightly more poleward initial track means that the NW turn occurs before the storm can reach Taiwan. AI models have been very good with track and large-scale patterns like this for a few years now, so I would lean in that direction for this forecast, but this will be something to monitor.

Figure 4: ECMWF forecast 500-hPa height and anomalies initialized at 06 UTC July 6, 2026, valid at 18 UTC July 7, 2026. Figure from www.tropicaltidbits.com.
Figure 5: EC-AIFS forecast 500-hPa height and anomalies initialized at 06 UTC July 6, 2026, valid at 18 UTC July 7, 2026. Figure from www.tropicaltidbits.com.

Even on AI models, however, there are some long-term pattern and track discrepancies that could affect how quickly the TC recurves and whether it dissipates over China or recurves towards Korea as it undergoes extratropical transition. For example, GraphCast shows ridging to the north of the TC (Figure 6) that allows it to continue inland into China, while FourCastNet-v3 shows weaker ridging and the TC recurves into the Korean Peninsula (Figure 7). This will be a medium-range trend to monitor.

Figure 6: GraphCast forecast 500-hPa height and anomalies initialized at 00 UTC July 6, 2026 and valid at 00 UTC July 11, 2026.
Figure 7: FourCastNet-v3 forecast 500-hPa height and anomalies initialized at 00 UTC July 6, 2026 and valid at 00 UTC July 11, 2026.

East Pacific

The East Pacific is quiet for now, but models continue to think activity will pick up as the month progresses (Figure 8).

Figure 8: Google DeepMind 50-member ensemble forecasts of TC tracks and intensity over the East Pacific initialized at 06 UTC July 6, 2026, valid out to 360h. Data from Google.

The most likely source for development will be the monsoon trough to the South of Mexico, where strong trade winds moving from the Caribbean will combine with weaker trades or westerlies to the south to produce an area of vorticity (Figure 9). Any development still looks to be over a week away, so this will be something to monitor.

Figure 9: AIFS forecast 10-meter wind and MSLP initialized at 00 UTC July 6, 2026 and valid at 00 UTC July 16, 2026.

Atlantic

The Atlantic remains very quiet as is typical during strong El Niño seasons and also consistent with July climatology. There are a few ensemble members with a vague signal for some development off the East Coast or in the far East Atlantic (Figure 10), but nothing worth diving into further for now.

Figure 10: Google DeepMind 50-member ensemble forecasts of TC tracks and intensity over the Atlantic initialized at 06 UTC July 6, 2026, valid out to 360h. Data from Google.

We’ll be back soon with updates on Bavi and other potential systems!