After Okinawa, the Models Stop Agreeing: What Dolphin's Track Split Means for China, Korea and Japan
Up to about 7 August, the forecast for Typhoon Dolphin is tight. After that, it comes apart.
This is the part of a forecast that gets flattened into a single line on a news graphic, and it is the part worth understanding, because the disagreement itself is information.
The setup: Dolphin has been tracking west, passing south of Chichijima on 4 August, and current guidance brings it near Amami Oshima and Okinawa around 7 August. Okinawa Island has about 1.4 million people.
Then it moves into the East China Sea, on or about 8 August, and slows down.
That is where the models split, and the split is wide.
What the models actually disagree about
The disagreement is not about whether Dolphin exists or where it is on Friday. It is about what steers it after the Ryukyus.
One group — the American GFS, the Japanese JGSM, and NAVGEM — has been dipping the system further south, some runs taking it south of Okinawa, before turning it sharply north later.
Another group — the European ECMWF and the GEFS ensemble mean — runs it southwestward instead, toward the coast of China, out around five days.
The broader consensus, including the AI-based models that have entered operational forecasting in the last few years, sits in a tighter envelope: west-northwest and decelerating.
Three plausible futures, then. A northward recurve toward Kyushu, the Korean peninsula and the Sea of Japan. A westward push toward Shanghai and Zhejiang. Or a slow grind west-northwest through the East China Sea that keeps everyone in the wind field for days without a clean landfall anywhere.
China's meteorological authority has said the picture should become clearer after 6 August, and has flagged Shanghai and Zhejiang as possibly affected. That is the correct way to phrase it at this range, and it is worth noticing that the agency closest to the outcome is the one being least specific.
Why forecasts fall apart at exactly this point
Because the storm is about to lose its steering.
A typhoon does not choose a direction. It is carried by the larger flow around it — principally the subtropical ridge, the belt of high pressure that sits over the western Pacific in summer. While that ridge is strong and well-defined, the storm runs along its edge and the track forecast is good.
In the East China Sea in early August, that ridge is often weak, broken, or being pushed around by mid-latitude troughs coming off the Asian continent. A storm that arrives there does not get handed cleanly from one steering feature to the next. It sits in slack flow.
Small differences in how a model handles the ridge produce large differences in where the storm ends up, because a slow storm in weak steering has time to drift.
That is also why the deceleration matters more than it sounds. A fast-moving system commits to a track. A slow one keeps its options open, and every extra hour of ambiguity multiplies the spread on day four and day five.
So the honest description of Dolphin after 8 August is not "heading for X". It is "in the East China Sea, slowing, with the ridge undecided".
What to do with a forecast that disagrees with itself
This is the practical part, and it inverts what most people do.
The instinct when models disagree is to wait for clarity before acting. That is backwards. A split forecast means the range of outcomes is wide, which means the chance that your particular location is affected in some way is higher, not lower.
If you have travel through Okinawa, Amami, Kyushu, Shanghai, or the Korean south coast between roughly 7 and 11 August, the correct action is to check your airline's waiver policy now, while the forecast is still uncertain. Typhoon waivers are usually published before the disruption, and they are easier to use before everyone else is calling.
Do not fixate on the landfall point. In a slow, broad, weakening system the landfall point is the least useful number available. Rain and wind arrive far from it, and the disruption to transport starts a day or more ahead of it.
Watch the wind field rather than the category. Guidance has gale-force winds spreading across roughly a thousand kilometres of the East China Sea within about five days, even as the peak intensity falls.
And follow one agency, the one responsible for where you are. Reading JMA, JTWC, the China National Meteorological Center and a news tracker simultaneously will give you four different numbers for the same storm and no clarity at all. That is a problem of measurement conventions, not disagreement about reality.
A correction from this site
Earlier this week we published a piece framed around Dolphin reaching Japan on 7 August and then Jiangsu on 9 August.
The first half has broadly held, with the caveat that the target has shifted south from Kyushu toward Okinawa and Amami. The second half was stated with more confidence than the guidance supported. Jiangsu is one scenario inside a spread that also contains a northward recurve and a slower grind through the East China Sea.
That is the failure mode this site keeps writing about, committed by this site: turning a probability envelope into a place and a date because a place and a date make a better headline.
Where to check this yourself
Japan Meteorological Agency: https://www.jma.go.jp/en/typh/
Korea Meteorological Administration: https://www.kma.go.kr/eng/weather/typoon/typhoon_05.jsp
Joint Typhoon Warning Center: https://www.metoc.navy.mil/jtwc/jtwc.html
Hong Kong Observatory: https://www.hko.gov.hk/en/index.html
China National Meteorological Center: http://www.nmc.cn/publish/typhoon/typhoon.html
JTWC publishes the model spread, which is the thing this article is about. JMA publishes the official forecast circle for Japan. Both are more useful than any single track line.
The short version
The forecast is tight through about 7 August and wide after it.
Three broad outcomes are live: a northward recurve, a westward push toward Shanghai and Zhejiang, and a slow west-northwest grind through the East China Sea.
The models split because the steering ridge weakens over the East China Sea and the storm slows down inside it.
And a disagreeing forecast is a reason to prepare earlier, not later. Waiting for the models to converge means acting at the same moment as everyone else.
Comments
Post a Comment