Typhoons and Rainstorms Past Help Hong Kong Forecasts Today

Wetter, Wilder Weather Events Loom with Warming World

You may find yourself on a Hong Kong beach, on a roasting hot day when the sun feels extra searing, the air cloying and superheated. Though there’s not a breath of wind, and the sea appears calm, surf builds along the shoreline, rises a metre or more, then surges up the beach. The water pauses as if the entire sea has just risen, then falls back. Another line of surf builds, and the water surges again.

If you know traditional local weather lore, you might figure there is a typhoon out to sea, and it may even be bound for Hong Kong. Your suspicion may be reinforced if clouds arrive by dusk, and are tinted purple that bathes the world in an eerie glow. Or nowadays, you could simply take out your phone, and check one of the various weather apps that show locations, predicted movements and intensities of typhoons and other storms.

*Purple glow at sunset, before arrival of Typhoon Mangkhut, 15 Sept 201*8

With super computers – and climate change, can we accurately forecast typhoons?

These apps feature a wealth of information, with real-time satellite and radar images, plus forecasts of winds and rain more than a week ahead. At times, someone might notice an app indicates a typhoon will impact Hong Kong in the coming days, and post an image that’s widely shared on social media – reflecting the widespread fascination with weather phenomena that can cause severe damage and loss of life, while also being oddly welcomed in some ways: typhoon party, anyone?

Looking at a typhoon track predicted by an immensely powerful computer – which might have a million or more processors, compared to two or four in your desktop machine^[i] – it’s easy to think this will prove definitive. But a track from another computer model may disagree, while the actual outcome could differ from either of them. The forecasts change over time, becoming especially reliable perhaps three days or less from landfall; yet for rainstorms, which can also wreak havoc, there may be significant issues with forecasts made a day or even hours in advance.

Yet we live in an era of supercomputers and AI, and there’s no shortage of experts investigating the weather, including in Hong Kong; with some remarkable advances in technologies noted by Lam Chiu-ying, director of Hong Kong Observatory from 2003 to 2009:

High-end mobile phones are probably faster and have memories bigger than the Cray computer acquired by the HK Observatory in 1999. Cray was then a “supercomputer”.
With the computer speed and memory nowadays coupled with an enormous amount of 3-d meteorological data derived from satellite observations, weather patterns can be computed days ahead.
Forecasters no longer have to work out future weather maps in their head.
Lam Chiu-ying, message to Martin Williams, 28 June 2024.

So is there hope for better foretelling and warning of potentially disastrous rainstorms and typhoons? And what of complications arising from climate change?

It turns out the answer requires looking back in time, especially to discover what’s possible.

*Powerful waves pounding southern Cheung Chau, as Typhoon Mangkhut heads for Hong Kong, early morning of 16 September 2018*

Hong Kong is prone to weather disasters

History shows that Hong Kong is prone to weather disasters that can cause significant loss of life, and have too often taken the territory by surprise.

“A fearful rain storm occurred at Hong Kong on May 29,” reported the Argus in Australia in June 1889. “The centre of the town is described as being simply a wreck.… Tramcars and tramlines were washed away…. Every street was flooded.”^[ii] There were hundreds of landslides on Hong Kong Island, and at least 27 people were killed.^[iii] Even today, this event holds several Hong Kong rainfall records including for 24-hour rainfall, though was not the deadliest of Hong Kong rainstorms. That dubious distinction is held by a succession of rainstorms that arrived over 16 to 18 June 1972, killing 150 people, 76 of whom died as a 13-storey apartment block on Kotewall Road toppled and smashed its way down a slope.^[iv]

Typhoons have proven even more calamitous, and have long afflicted Hong Kong and nearby areas. The Hong Kong Observatory originated as a result of a typhoon that struck on 22 September 1874^[v]^[vi], killing around 2,000 people in Hong Kong – which then comprised only Hong Kong Island and the Kowloon peninsula. According to a history of the Observatory, this showed “an urgent need for an organisation that could predict the arrival of storms and issue warnings,” leading to its establishment in 1884.

As recounted in a paper titled “Uncertain skies: Forecasting typhoons in Hong Kong ca. 1874-1906” by Fiona Williamson of Singapore Management University, the Hong Kong Observatory soon became involved in developing forecasting techniques, particularly for typhoons. The first director, William Doberck, published a pamphlet on typhoons, and Williamson reports that towards the end of the century, there were heightened public expectations regarding weather forecasts.

A typhoon that arrived with little warning on 18 September 1906 dashed public hopes regarding typhoon forecasts. Though they had anticipated little but a few thunderstorms, Observatory meteorologists noticed an abrupt air pressure change in the morning, and hoisted the black drum – a tropical cyclone warning – at 8am. Just 40 minutes later, the typhoon gun sounded over the harbour, sending panicked sampan and junk operators fleeing for safety. For many, it was too late; the typhoon’s onslaught, lasting under two hours, killed at least 10,000 people, many of them on boats that were sunk and smashed in Victoria Harbour.^[vii]

Meteorology is not an exact science

The next year, a government report on the tragic typhoon quoted Doberck: “Meteorology is not an exact science. Nothing can be predicted with certainty.”

Doberck’s statement has since held true, including for Hong Kong’s strongest recorded typhoon, which rampaged across the territory on 2 September 1937. There was some warning, with the Number 1 signal hoisted on 31 August, yet it was thought the storm would pass to the south of Hong Kong, until a course change was noticed the following afternoon and the Number 8 signal was raised^[viii]. The situation was ripe for disaster, with Time later reporting the harbour was “unusually jampacked with shipping”, including many vessels that had sought refuge following Japan’s invasion of Shanghai.^[ix] Plus, the hurricane force winds arrived after midnight, coinciding with high tide.

As noted by a pithy report in the Waikato Times on 3 September 1937, “The waterfront is strewn with large and small craft which have been blown ashore. Among them are 21 sea-going vessels… The roads are littered with the debris of houses and motor-cars.”^[x] A Post reporter headed to Tai Po, and described “scenes of desolation” including bodies still strewn over the streets and the wreckage of buildings in Tai Po Market, as a result of a storm surge that swept in like a tidal wave.^[xi] Around 11,000 people were killed.

*Road to Cheung Chau helipad wrecked by Typhoon Mangkhut storm surge waves on 17 September 2018*

The Observatory’s former history man, Shun Chi-ming

While the Hong Kong Observatory stored documents on these and other natural disasters, they received little attention until weather man doubling as history man Shun Chi-ming became director.

Shun tells of becoming fascinated by physics as a schoolboy, and finding science rather than philosophy offered the most concrete approach to understanding the world. While studying physics at the University of Hong Kong, he took a course on atmospheric physics. “I didn’t know that with physics and maths you could earn a job in Hong Kong Observatory,” he says. “You can apply theoretical studies to real life situations, and save some lives.” On graduating, he applied to the Observatory, and was accepted, becoming a scientific officer.

After assignments including radiation monitoring and seismology, and a forecasting course for senior scientific officers, Shun had a six-week stint in the UK with the European Centre for Medium-Range Weather Forecasts – which today operates the world’s premier computer forecast model. “It was around 1990, no one trusted the computer models; the results were very coarse,” recalls Shun. “I helped with international cooperation, to provide intensive observations over east Asia – and we found that with more data, the model performed better.”

Prior to the 1990s, the best method for forecasting a typhoon depended on measuring the speed it was moving, along with assessing the climatology – the overall weather pattern, with consideration of the time of year. This, plus comparison with previous storms with similar weather patterns, helped predict its future movement.

“Even with AI today, there is basically this analogue process, supported by sophisticated deep learning,” explains Shun. “The basic principle hasn’t changed, but there’s a big dataset.”

Shun became director of the Hong Kong Observatory in 2011, and soon became interested in historical storms. “I went to a celebration in Macau, and there was an exhibition on the 1874 typhoon, in which eight percent of the population died,” he says. Shun then learned the typhoon had killed over 2000 people in Hong Kong, and decided to study more about the past. Reading documents on storms such as in 1906 and 1937, he became concerned about the potential for further storm surges.

There had been no major surge since 颱風旺達, which had killed at least 130 people in 1962^[xii], yet as further typhoons loomed, Shun was always alert to the possibilities, and drew on his historical knowledge.

*Tong Fuk Beach, south Lantau, damaged – and with beach sand eroded away – by storm surge from Typhoon Hagupit in September 2008*

Historical knowledge helped predict Mangkhut’s impacts

When Typhoon Hato approached in 2017, Shun had to decide if it could be like Wanda, or more akin to Hagupit – which in September 2008 passed south of Hong Kong, causing gales and a storm surge that mostly affected southern Lantau and Cheung Chau. “I looked at 1937 and 1936 typhoons, which were of similar strength, but the 1936 storm was perhaps 20 to 30 kilometres to the south, not causing a destructive storm surge,” Shun says. “The 1936 typhoon track overlapped the track of Hato, and was further south than Wanda, so I decided it would be more like Hagupit.” Though Shun issued the T10 signal, and hurricane-force winds were accompanied by a substantial storm surge^[xiii], Hong Kong was indeed spared a Wanda-like assault.

A year later, Shun and colleagues became far more concerned as Super Typhoon Mangkhut formed over the western North Pacific, and took aim at Hong Kong. Their computer model indicated the potential for an immense storm surge, raising water levels by over five metres in Victoria Harbour, and an additional two metres in Tolo Harbour: no one in the observatory had experienced anything like this. Looking for historical comparisons, Shun figured it would be similar to the 1937 typhoon, or Wanda. “It was stressful making this decision – once I said it could be like Wanda, there was no turning around,” he says.

The Observatory started alerting government departments a week before Mangkhut’s predicted arrival, and held press conferences three and two days in advance. “During Mangkut, I worked over 40 hours, with just one or two hours sleep,” says Shun. “It was really exciting; I didn’t want to miss it.” The enhanced observatory staff included two forecasters rather than one, and a social media team that disseminated information, and showed Shun examples of images coming in as the surge flooded homes in Sai Kung, giant waves pounded Heng Fa Tsuen housing on eastern Hong Kong Island.

*Tree felled by Typhoon Hagupit on Cheung Chau, 16 November 2018*

Though over 458 people were injured, tens of thousands of trees were downed^[xiv], and the 3.88 metre surge in Victoria Harbour was akin to that during Wanda, Shun considers Hong Kong got lucky – helped by a brush with north Luzon somewhat reducing Mangkhut’s intensity, and low tide coinciding with the storm’s closest approach.

Shun retired from the Observatory in 2020, and recalls issuing typhoon signals as being stressful for the director. “You need to be careful in striking the right balance,” he says. “It’s a million dollar or even multi-billion dollar decision.” Though he issued the T8 signal for Typhoon Linfa in 2015, when “nothing happened”, he has no regrets, noting: “It’s better to be safe than sorry.”

Currently, Shun is helping the Hong Kong University of Science and Technology establish a climate centre focused on adaptation and resilience, such as preparing for more extreme typhoons as global temperatures continue rising. “We’re gathering historical climate information to develop a detailed model, such as showing the places in Hong Kong with the heaviest rain, and highest temperatures,” he says.

Even forecasting numbers of typhoons per year is tough

Given that typhoons are powered by warm ocean waters, it might seem likely that as the Pacific waters become warmer, the number of typhoons per year will increase. Yet the true picture is not so simple.

Even forecasting the numbers of typhoons in a given year remains far from an exact science. In May this year, StormGeo – which focuses on weather services for businesses – anticipated below normal typhoon activity in 2024, with perhaps 12 in all. The company’s Senior Meteorologist, Nancy Lin, cautions against taking this number as more than a curiosity, noting that, “The primary objective of predicting the number of typhoons or hurricanes is to facilitate proactive planning, especially for weather-sensitive industries like shipping and offshore wind.”

Johnny Chan, an Emeritus Professor at the City University of Hong Kong, has spent four decades studying extreme weather, and is surely the world’s leading expert on typhoon science. “El Nino is the most important influence on annual numbers of typhoons, but doesn’t explain everything,” he says. “Other factors include sea surface temperatures in the Indian Ocean, and decadal changes.” He has helped develop a system that he rates “quite good” for predicting numbers of typhoons making landfall on the South China coast, but tends to underestimate tallies for Japan and Korea.

Regarding climate change and numbers of typhoons, Chan says, “If water temperature was the only factor, the frequency would increase; but there are other things happening.” Most studies – including by Chan – suggest a reduced tendency for clusters of clouds to begin rotating en masse, as required for the very earliest stages of typhoons. However, once they begin forming, typhoons are likely to be more intense than in previous decades.

*Smashed windows of holiday flat on Cheung Chau, battered by Typhoon Hagupit wind and waves on 16 September 2018*

Forecasting typhoon tracks has improved but mysteries swirl regarding intensities

Even as nascent storms are detected, computer models set to work on predicting their movements, and they can be remarkably accurate. “If you look at track forecasting, the errors have decreased substantially in the last 20 years,” says Chan. “But there are some tracks that differ significantly from what is observed, and we need studies on why such predictions are so bad.”

Of course, not everyone considers an approaching typhoon means it is time to seek safety. Even in 1884, the Hong Kong Telegraph reported, “The two typhoon guns brought out a considerable number of enterprising sightseers who braved all the discomforts of the roaring gale and blinding rain for the purpose of witnessing Nature in one of her fiercest moods,” – and in recent years social media posts have shown people likewise checking out wind and waves. There are even professional storm chasers, including Japan-based Briton James Reynolds, whose career centres on documenting typhoons across east Asia.

For Reynolds, a successful chase requires getting into place ahead of the storm, near or exactly where the eye will pass. “I gather all the raw data and information from different models, and form my own opinion,” he says, echoing Chan in noting, “Over the years, track forecasting has improved tremendously.”

Even so, Doberck’s comment about meteorology being inexact holds true. “The big mystery is the intensity forecast,” says Reynolds. “When I’m chasing in the formative stage, I have no idea how strong a storm will be – I go in, and it could be category 1, to category 5 [as per the Saffir-Simpson Hurricane Wind Scale]. It’s a crap shoot.”

Typhoon strengths affected by factors including warmth over Tibetan Plateau

Johnny Chan sees similar issues, remarking, “Intensity forecasts haven’t really improved for about 20 years – especially for typhoons that are about to make landfall.” The latter point is especially noteworthy given there has been a recent increase in numbers of typhoons forming and rapidly intensifying near coasts, including in the South China Sea.

“That’s a worrying thing for the south China region,” says Chan. “We don’t know all the physics behind rapid intensification, and models can’t capture the situation very well.”

Chan has gleaned some ideas about this rapid intensification from studying cases such as Typhoon Hato, which bypassed Hong Kong while gaining strength before severely impacting Macau. “There was relatively little cloud cover, and hot, sinking air ahead of the storm that warmed the sea surface, which may have boosted the storm,” says Chan. With Mangkhut, by contrast, there were extensive clouds coupled with an approach that was mainly at nighttime, which may account for its intensity changing little before landfall.

Early this year, Chan and co-authors published research that linked stronger typhoons to warmer temperatures over the Tibetan Plateau, as a result of decreased vertical wind shear (Increasing tropical cyclone intensity in the western North Pacific partly driven by warming Tibetan Plateau). “A typhoon is a three-dimensional thing, rising very high in the atmosphere, and vertical wind shear is the change of regional wind speed and direction with height,” explains Chan. “With little shear, the typhoon is like a vertical cylinder, and can be very strong. But with stronger vertical shear, the cylinder becomes tilted, and there’s less likelihood of high intensity.”

While the decrease in shear had been reported by others, this was the first study to explain why it is occurring, as winds from the warmer Tibetan Plateau modify the atmospheric circulation over the Pacific.

Heavier downpours with warmer, dirtier air

Chan also says the amount of rain from typhoons has also increased, particularly in coastal cities. This is partly as a warmer atmosphere can hold more moisture, but also results from warm air rising over built-up areas that are heated by the sun, and have sources of heat like air conditioners and automobiles. Plus, pollution results in more airborne particles that water condenses around, forming initially smaller drops that rise further before agglomerating, freezing and falling as hail and intense rain.

This conclusion tallies with research by Kerry Emanuel – American professor of meteorology currently working at the Massachusetts Institute of Technology in Cambridge – and colleagues:

We have looked at tropical cyclones in the far western Pacific and the South China Sea. The main projected effect of climate change in this region is to increase typhoon rainfall. For example, we think the probability of a typhoon producing a meter of rain in Hong Kong increases from once in about 300 years to once in about 80 years. We do not see much change in wind intensity.
Typhoons and hurricanes almost always decay quite rapidly as they move inland. Although there is a paper suggesting that “brown oceans: can keep these storms going, the physics suggests that one needs exceptionally hot, dry soils that suddenly release heat when they are wetted by the first rains of a typhoon. This happens prominently in tropical Australia, but not much in China.
We have also looked at whether the duration of typhoon rains at Hong Kong changes. We find that most of the 9 CMIP6 models we downscaled indeed show increased durations of heavy rains at Hong Kong, a few show the opposite.
Kerry Emanuel, email to Martin Williams, 19 June 2024

The warmer, more polluted atmosphere cited by Chan also explains why rainstorms are tending towards more severe, with the hourly rainfall record mostly below or little over 100 millimetres until 2008, when it reached 145.5 millimetres, with a further jump to 158.1 millimetres on 7 September last year. While confident of this trend, Chan says rainstorm events are very difficult to predict, partly as a rain formation area is like a weak cold front that can migrate perhaps 100 kilometres to the north or south, in which case Hong Kong would receive little rain.

*Rainstorm arriving over Hong Kong, from Cheung Chau*

Hong Kong needs to prepare for storm surges

Even as wetter and wilder weather events lie ahead, Shun Chi-ming believes it is still useful to look back in time. “Another historic typhoon I’m interested in was in 1862, hitting Guangzhou and the west part of the Pearl River delta,” he says. “While there were different estimates of casualties, perhaps 80,000 people were killed. It’s something for the Greater Bay Area to take a look at.”

Now, even Mangkhut is part of history, and Shun tells of preaching to government officials that it could have been even stronger, with a higher storm surge. “I’ve highlighted that it’s a myth to believe underground storage can take enough water to mitigate a storm surge,” he says – instead, it’s best to prepare as much as possible for inundation by the sea.

As we go to press, the monsoon rains have mostly shifted to the north of Hong Kong, the west Pacific is peaceful. Yet any day now, thunderstorms may begin clustering over the Pacific or the South China Sea, and the cluster may begin spinning, gently at first, then faster, as a central vortex develops where winds begin to rage. Soon, those waves may be rolling up beaches from a calm sea, with intense heat building. A typhoon could be inbound for Hong Kong; and another historic weather event might be underway.

Written for the South China Morning Post Magazine.

^[i] https://www.ecmwf.int/en/about/media-centre/focus/2022/fact-sheet-supercomputing-ecmwf#:~:text=While%20an%20average%20desktop%20PC,million%20billion%20calculations%20per%20second).

^[ii]