While Hong Kong is sometimes hit by typhoons, predicting them in advance is tricky. Yet this doesn’t stop social media and even mainstream media posts that latch on to occasional results of computer forecast models that suggest potential typhoons, even while other results at the same time show little or nothing special.
For instance, here are two forecasts from two computer models – by ECMWF and GFS; both predictions made on 15 July, and both for 21 July 2024:
Never mind it being days ahead, with no real agreements between forecast models, even prior to the above there were media articles suggesting a typhoon could hit Hong Kong around 21 July. Yet as it turned out, only a minor low pressure system developed; so this was a classic case of crying wolf
At the top of this page is a screenshot from one of these articles – in the South China Morning Post. This was published on Friday 12 July; led to it being shared on social media, and to some people it appeared the “Sunday” could mean a typhoon was about to hit within a couple of days! Instead, a look at the Observatory blog revealed the potential typhoon was for Sunday 21 July – and even then it was only a somewhat faint possibility, appearing in a computer model run or two, but with very little agreement.
Media website HK01 published a longer article, noting that a mainland China weather blogger was anticipating perhaps four typhoons including super typhoon. This led to Shun Chi-ming, a former director of the Hong Kong Observatory, being told by a restaurant waiter that there would be four typhoons; he promptly responded to the over excited forecasts with a Facebook post debunking the wilder claims [in Chinese]. HKO1 has since revised its article; and as I write on 18 July, there is nothing about an actual typhoon being imminent.
And early on 22 July, here’s a weather chart; no super typhoon or even typhoon around, just a tropical storm at Hainan; while all is calm weather-wise in Hong Kong:
So while it’s kind of exciting to spot a potential typhoon in computer model forecasts, it’s also important to consider if the result is an outlier, and that other forecast outcomes are more likely – and not publish “possible typhoon” headlines when the possibility remains remote.
From my experience of “watching” typhoon forecasts, the probabilities of an actual typhoon rise as forecast models better agree with one another; this generally requires at least having a tropical storm developing. And, of course, the forecasts tend to become more accurate as the time to a storm’s possible arrival decreases.