Severe Weather in HK

E W L Ginn, S C Tai, O Lee, K L Ho, C H Chow and T H Chow

Severe Weather in Hong Kong

1. Introduction

Hong Kong is situated in a region where both tropical and mid-latitude weather phenomena are experienced. Tropical weather systems like tropical cyclones and troughs of low pressure are common in summer. Mid-latitude systems like cold fronts brought cold surges in winter. These systems, together with the monsoons (northeast and southwest in winter and summer respectively) give rise to a wide variety of weather in Hong Kong. To accurately predict the occurrence and intensity of these events makes weather forecasting here a most challenging job.

This package introduces to you four severe weather events in Hong Kong. It comprises short descriptions of the events in text together with weather maps, satellite pictures and sequences of radar animation.

2. Weather Radar

RADAR  is an acronym for : RAdio Detection And Ranging. A weather radar detects rain in the atmosphere by sending out pulses of microwave from its scanning antenna. The antenna acts both as a transmitter and a receiver. When a radar pulse encounters rain drops, part of its energy is reflected back to the antenna. The time delay in the returned signal, that is, the echo, gives information on the distance of the rain area from the radar. Rain intensity is calculated from this distance and the strength of the received signal as well as assumptions about raindrop size.

Fig. 1 Schematic diagram showing how a weather radar works

Weather radars have become an indispensable part of weather forecasting in Hong Kong in the past 30 years. As early as 1959, the Hong Kong Observatory installed at Tate's Cairn its first weather radar. Over the years, weather radar helped forecasters in monitoring rain development and movement. The data collected forms a solid base for research into severe weather phenomena associated with heavy rain.

In its normal mode of operation, the Observatory's  Radar at Tate's Cairn scans the atmosphere at 12 discrete elevations from 0.5 degree to 35 degrees above the horizon every six minutes. At each of these elevations the antenna scans through a full circle. The data collected during these 12 scans are then combined to form a  "volume scan" which provides a 3-dimensional database on the rain areas within around 250 kilometres from the radar station. A computer workstation then processes the volume scan data and generates various types of radar pictures for use by weather forecasters.

The most popular type of radar pictures used by weather forecasters are the Constant Altitude Plan Position Indicator (CAPPI) pictures, which show rain areas in a horizontal cut of the atmosphere at a fixed height above ground. Typically, weather forecasters use the CAPPI pictures at 3 km height for daily forecasting purposes. In this package, all radar pictures are also CAPPI pictures at 3 km height.

3. Four severe weather cases :

Case 1. Hailstorm (19 April 1995)

Radar animation  from 4:54 p.m. (1654H) to 8:54 p.m. (2054H)

Hail is largely a mid-latitude phenomenon. However, in spite of its relatively low latitude (22.3 degrees N), Hong Kong has experienced 25 occasions of hail during the past 30 years (1966 to 1996). Over 80 percent of these events took place in March and April.

Intense convection is triggered off in an unstable atmosphere (e.g. near a trough of low pressure). As moist air rises in the atmosphere, it gradually cools down and eventually moisture condenses into water droplets. As the parcel of moist air rises further above the freezing level, some of the water droplets will condense on solid particles suspending in the air and form small ice pellets. Other water droplets then condense on the surface of the small ice pellets, which develop into larger ice pellets. These ice pellets can grow to the size of hailstones if there is a strong and persistent uplifting airstream to hold the ice pellets up in the air above the freezing level for long enough time.

A hailstorm in Hong Kong occurred on 19 April 1995 during an episode when an active trough of low pressure lying close to the south China coast (see Fig. 2a) brought thunderstorms to Hong Kong.
 

Fig. 2a Surface weather map at 2 a.m. 19 April 1995.
 

Fig. 2b Part of the radar picture of 6:30 p.m. (1830H).

The radar animation sequence showed clearly the echoes associated with the hailstorm about 60 kilometres northwest of the radar station at about 5:06 p.m. (1706H). The hailstorm came close to Tai Po at around 6 p.m. and brought hail to the area. Note the 'hook' shape echo in the radar picture of 6:30 p.m. (1830H), as shown in Fig. 2b, which is a typical signature of hailstorms.

According to press reports, hail started to affect areas west of Guangzhou in the early afternoon, and by the evening the storm began to hit the northern part of Hong Kong. Coinciding with the passage of the hailstorm, there were reports of hailstones 1 cm in diameter over a 10 km-wide area near Tai Po. A raingauge near the area also recorded more than 80 mm of rain during the passage of the storm. The daily rainfall distribution map of Hong Kong for 19 April 1995 is shown in Fig. 2c.

Fig. 2c 24-hour rainfall distribution map for 19 April 1995.

Case 2. Severe Tropical Storm Helen (12 August 1995)

Radar animation from 00:12 a.m. (0012H) to 5:42 p.m. (1742H)

Helen formed as a Tropical Depression over the western North Pacific about 1200 km east of Manila on 7 August 1995. Tracking northwestwards, it intensified into a Tropical Storm on 9 August and entered the South China Sea one day later. Helen intensified further into a Severe Tropical Storm about 380 km south-southeast of Hong Kong on 11 August. The No. 8 NORTHEAST Gale or Storm Signal was hoisted in that evening as Helen came within 150 km south-southeast of Hong Kong. It made landfall about 60 km northeast of Hong Kong the following morning and then dissipated over land in the evening. Fig. 3a  is a plot of Helen's  track while Fig. 3b is the surface weather map at 2 a.m. on 12 August 1995 showing the surface pressure and wind pattern associated with Helen. Fig. 3c is a satellite picture taken at roughly the same time, showing the deep convection near the centre of Helen and its surrounding spiral cloud bands.

Fig. 3a Track of Severe Tropical Storm Helen : 7 - 12 August 1995.

The radar animation sequence showed that over the sea, the eye of Helen (an echo free area at the geometric centre of the tropical cyclone surrounded by spiral rainbands) was initially rather difficult to locate but became more well-defined starting from about 4:42 a.m. (0442H) on 12 August. As Helen crossed the coast and moved inland after around 9 a.m., the supply of moisture from the sea was cut off and the cyclone started to weaken. The centre then became less distinct again.

Fig. 3b Surface weather map at 2 a.m. on 12 August 1995

Fig. 3c A satellite picture taken at around 2 a.m. on 12 August 1995

Bearing in mind that air rotates cyclonically (anti-clockwise in the northern hemisphere) around a tropical cyclone and that the pressure decreases as one approaches the centre, it is interesting to note how the local winds and pressure changed during the passage of Helen. Fig. 3d gives the time series of the winds and pressure recorded at King's Park in Kowloon from 4:00 a.m. to 10:00 a.m. on 12 August. Initially when Helen was to our south, winds over the territory were typically from the northeast. As the cyclone came closer to the coast east of Hong Kong, winds gradually turned northwesterly and then westerly. Winds further turned and became southwesterly when Helen was over land to our northeast. The pressure fell steadily as Helen approached the territory and reached a minimum between 6 and 7 a.m., when she was closest to Hong Kong. The pressure gradually increased as Helen moved away and weakened.

During its passage, the high winds of Helen brought down trees and scaffoldings in many parts of the territory. Helen also brought heavy downpours to Hong Kong, causing widespread flooding and several landslips. Fig. 3e shows the daily rainfall brought by Helen on 12 August. Three people were buried to death in the serious landslips in Chai Wan and Aberdeen on 13 August. After making landfall in Guangdong Helen claimed 23 lives and about 54,000 houses were damaged or destroyed.

Fig. 3e 24-hour rainfall distribution map for 12 August 1995.

Case 3. Typhoon Kent (31 August 1995) 

Radar animation from 7:00 a.m. (0700H) to 11:12 p.m. (2312H)

Kent formed as Tropical Depression around 670 km east of Manila on 26 August 1995. By 30 August, Kent intensified into a Typhoon and entered the South China Sea. The No. 8 NORTHWEST Gale or Storm Signal was hoisted at 1 p.m. on 31 August when Kent came within 130 km east-northeast of Hong Kong. Kent soon made landfall over eastern Guangdong and weakened into an area of low pressure on 1 September.

As shown in the radar pictures sequence, Kent's centre could be identified as it entered the radar screen from the east at around 9:00 a.m. on 31 August. Just like what happened to Helen, Kent's centre became difficult to identify and track after it made landfall to the east of Hong Kong.

Fig. 4a is the track of Kent while Fig. 4b is a surface weather map at 8 a.m. on 30 August. Kent's path did not meander like Helen and it moved at a steady speed of around 22 kilometres per hour during the hoisting of No. 8 Signal in Hong Kong.

Kent is a good example to illustrate the theory that a tropical cyclone can be thought of as a ball moving along a stream of water. The water stream is the background environmental flow which steers the ball. Similarly, a tropical cyclone's movement is largely determined by the steering flow in the atmosphere.

Fig. 4a Track of Typhoon Kent : 25 August - 1 September 1995.

Fig. 4b Surface weather map at 8 a.m. 30 August 1995.

Fig. 4c Streamline chart of wind flow at 500 hPa, at 8 a.m., 30 August 1995

Fig. 4d  A satellite picture taken at around 9 a.m. on 31 August 1995.

Kent winds uprooted many trees and its heavy rain caused flooding and landslips in the territory, especially in Mid-levels, Tuen Mun, and Sheung Shui. More serious damage was inflicted by Kent in Guangdong where at least 50 people were killed and more than 40,000 houses collapsed or damaged.

Case 4. Rainstorm (29 March 1996)

Radar animation from 1:00 a.m. (0100H) to 12:00 noon (1200H)

An active trough of low pressure affected the coast of Guangdong from the northwest on 29 March 1996 (see Fig. 5a). Fig 5b is a satellite picture taken at around the same time, showing an extensive cloud band lying nearly over the trough. As shown in the radar animation sequence, a northeast-southwest oriented rainband quickly swept across the coast of Guangdong during the morning.

Fig. 5a Surface weather map at 2 a.m. on 29 March 1996.

Fig. 5b A satellite picture taken at around 2 a.m. on 29 March 1996

Embedded in the rainband were numerous rapidly developing and dissipating convective cells which were much smaller in size. These cells sometimes merge and develop into a line of thunderstorms and brought abrupt changes in wind direction in squall lines in addition to heavy rain. The squall line in this case is related to the active trough along the coast of south China. Strong descending airstreams associated with the intense convective cells gave rise to high winds on hitting the ground. Fig. 5c is the time series of wind data (averaged over 10 minutes) at Cheung Chau from 7:00 a.m. to 10:00 a.m. on 29 March 1996. Note the picking up of wind speed and change in wind direction from southeasterlies to northwesterlies when an intense convective cell crossed Cheung Chau at around 8:00 a.m. followed by an even stronger one at around 9:00 a.m.

During this rainstorm, the thunderstorm and flood warnings remained effective for over 4 hours. Rainfall in excess of 50 millimetres are recorded on 29 March (see Fig. 5d). More than 100 shops were flooded in Mong Kok and air traffic was once affected.

Fig. 5d 24-hour rainfall distribution map for 29 March 1996.

4. Image interpretation - What do the Colours mean ?

Radar provides an estimate of the rainfall intensity of a rain area. In the radar pictures of this package, different rainfall intensities are represented by different colours :

The table below shows the meaning of colours in Radar Images:

Colour of displayed echoes	Legend	Rainfall Rate (mm/hour)
	Yellow	0.15 to 2
	Green	2 to 5
	Light Blue	5 to 10
	Grey Blue	10 to 20
	Blue	20 to 30
	Dark Blue	30 to 50
	Purple	50 to 100
	Light Purple	100 or above

We should, however, bear in mind a number of points when interpreting the colourful pictures.

Firstly, the radar beam is a conical ray and spreads out as it goes further in the atmosphere. At a distance of 50 km from the radar station, for example, the radar beam is already about 1.5 km wide, and the radar will be sampling a fan-shaped volume of about one cubic kilometre in the atmosphere! Rainfall intensity at all points in this sample volume will be averaged and represented by one colour in the radar picture.

Secondly, in converting the strength of the received radar signal into rainfall rate, we have made the assumption that the raindrop sizes within the sample volume follow a particular distribution formula, but in fact, the raindrop size distribution may be significantly different in different weather systems.

Thirdly, the radar pictures show a horizontal cut of the atmosphere at around 3 kilometres above the ground. The rainfall rates as measured at this height may be quite different from what we observe at the ground level.