Reprint 445

Roles and Functions of National Meteorological

and Hydrological Services - a Hong Kong Case Study

H.K. Lam

Regional Seminar on Meteorological Services:

Opportunities and Challenges in the Twenty-first Century,

Seoul, Republic of Korea, 17-18 September 2000

1

Roles and Functions ofNational Meteorological and Hydrological Services

- a Hong Kong Case Study

Dr. H.K. LamDirector

Hong Kong Observatory134A Nathan Road, Kowloon,

Hong Kong, China

Presented at the

REGIONAL SEMINAR ON METEOROLOGICAL SERVICES:OPPORTUNITIES AND CHALLENGES IN THE TWENTY-FIRST CENTURY

(Seoul, Republic of Korea, 17 to 18 September 2000)

1. INTRODUCTION

Hong Kong is a compact cosmopolitan city of about 1,000 squarekilometers in area with a population of about 6.5 million (Figure 1). It is situatedalong the coast of southern China and is affected by a diversity of weather systems.Historically, extensive weather-related casualties, both on land and at sea, weremostly caused by high winds and storm surge during the close passage of tropicalcyclones. But through the years, coastal engineering works and better structuraldesigns have gone a long way in mitigating these disasters. By the 1990s, the majorperennial weather hazards to be tackled are those associated with severe rainstormsand prolonged heavy rain. These result in landslips on unstable slopes, urbanflooding in localized areas with poor drainage, and extensive flooding in the low-lying plain over the northern part of Hong Kong.

Hong Kong is also a very vibrant and energetic city. It is always on themove and hardly has a moment of rest. Through the years, the society as a wholehas continued to evolve. The way of life changes and public expectation is neverconstant. In the context of social and economic development, weather warningsand services are no longer confined to the realm of public safety. They now play apivotal role in ensuring or even improving the quality of life as well. Standards andcontents of meteorological services have to keep pace with the changing needswithin the local environment. And on the global scale, information technology isleaving its mark in all walks of life. Our modes of operation and methods of

2

delivery are inevitably affected. In many ways, it acts as a spur for themodernization of meteorological services. It is no longer a matter of finding roomfor improvement; there is just no room left for not improving!

At the Hong Kong Observatory, we are keenly aware of the need to adopt aproactive and progressive stance in a continuous effort to upgrade our forecastingand information services. I shall review some of the major efforts undertaken bythe Observatory in recent years - the formulation of forecasting techniques gearedtowards the prediction of mesoscale systems and rainstorms, the utilization ofinformation technology in service provision, and the technical systems developedin support of major infrastructure projects such as the new airport in Hong Kong.

In the light of the Hong Kong experience, I shall attempt to extend some ofthe ideas on technical and service development to a more regional, or even global,situation. By doing so, it is hoped that we may be able to deal effectively with themany challenges facing the NMHSs in the new millennium through a programmeof strategic collaboration.

2. THE HONG KONG EXPERIENCE - TECHNICAL DEVELOPMENT

Significant rain associated with various weather systems can affect HongKong for most part of the year. The most critical period, however, is during theearly summer months of May and June when the monsoon troughs are active oversouthern China. Short-range forecast of the evolution of intense convective cellsembedded within these systems, from mesoscale downwards, remains a majorchallenge. The highly uneven rainfall distribution in many rainstorm cases,exacerbated by the hilly terrain and tortuous coastline of Hong Kong, rendersquantitative rainfall forecasts, even for only a few hours, extremely difficult. Todeal with the problem, we have to look at the available tools in state-of-the-artscience and technology and to adopt a workable development strategy applicable tothe local situation.

In a world of limited resources and knowledge, particularly as observationand forecasting of cloud-scale rain systems are still very much in the realm ofresearch investigation, a practical option is to pursue an engineering approach indealing with operational problems relating to the prediction and warning ofrainstorms. First of all, we need input data from a dense local observation network.Then in search of the forecast solution, a high-resolution numerical model and atailor-made nowcasting system are to be our toolboxes.

3

2.1 Observational Systems

2.1.1 Automatic Weather Observing Network

In Hong Kong, a dense network of over 70 automatic weather observingstations comprising of rain-gauges, anemometer stations, wind profilers, andautomatic weather stations (AWS, measuring wind, temperature and humidity)provides minute-by-minute real time data for forecast operation. The network isdesigned to be maintenance-free by using solar power and incorporating the latesttechnology in lightning protection and self-diagnostic software. With this network,severe weather systems on the mesoscale and local scale can be readily resolved.The network will be further expanded in the coming years in collaboration withneighbouring services in Guangdong and Macau to cover the entire Pearl Riverdelta in southern China.

2.1.2 Doppler Radars

The Observatory operates two Doppler radars for general weathersurveillance: one at Tate's Cairn and another one at Tai Mo Shan. By networkingthe two radars together and using advanced data processing software, forecastersare now provided with new radar products, such as dual-Doppler winds (Figure 2),composite reflectivity radar images, 3-dimensional display of storm structures(Figure 3), as well as forecast radar products based on user-defined advectionvectors or forecast winds from NWP models.

2.1.3 Aviation Systems

At the Hong Kong International Airport in Chek Lap Kok, the Observatoryoperates a dense network of weather sensors in addition to conventionalobservation of pressure, temperatures and rainfall (Figure 4). Minute-by-minutedata are collected in support of aircraft operation.

Thunderstorm-induced windshear and microburst events are monitored by aTerminal Doppler Weather Radar (TDWR) situated about 12 kilometres from theairport. The Radar commands an entire view of both runways, and the approachand departure areas.

Terrain-induced windshear events are determined using a network ofautomatic weather stations to monitoring surface weather conditions in themountainous terrain near the airport. A Windshear and Turbulence WarningSystem (WTWS) issues coherent alerts to air traffic controllers after reconcilingdata from all sources. Based on investigation of windshear events, the windsheardetection algorithm is continually refined.

4

The Observatory will also acquire a Light Detection and Ranging (LIDAR)system to enhance detection of low level windshear and turbulence at the airportunder non-rainy situations. The system makes use of the Doppler effect of laserlight to detect movement of particulates in the atmosphere.

2.2 Forecast Systems

2.2.1 Numerical Modeling

Driven by the need to forecast mesoscale convective systems for a smallarea like Hong Kong, much effort has been expended to develop a regional high-resolution numerical model. In 1997, with the kind support of the JapanMeteorological Agency (JMA), a Regional Spectral Model (RSM) was adapted forlocal use.

The model currently operates with two domains. An outer domain has 60-km horizontal resolution and 20 vertical levels. An inner domain is nested with20-km horizontal resolution and 36 vertical levels. Besides conventional GTS data,the model ingests digital cloud data from the GMS for moisture bogussing. Tominimize the spin-up problem, hourly rainfall amounts based on combined analysisof radar and raingauge data are incorporated through physical initialization.Whenever a tropical cyclone is in the domain, an asymmetric typhoon bogussingtechnique is used to improve track forecasts. Asynoptic data, such as hourlyobservations, will have more and more impact on model performance as thefrequency of running the analysis/forecast cycle is increased. We are working withour neighbours to extend the hourly AWS coverage. The latest ATOVS datadisseminated from the NOAA satellite are incorporated to obtain derivedtemperature and humidity data for more pressure levels. The model is currentlyrunning on a CRAY SV1 super-computer with 16 CPUs and a peak performance ofabout 19.2 GFLOPS.

Post-processing of model outputs utilizes regression techniques andKalman filtering to improve point forecasts. Summarized weather map showingfine, cloudy and rainy areas, time series of local weather elements, automatedworded weather forecasts and warnings (Figure 5) are also generated to assist theforecaster to assimilate and interpret the model results for operational reference.

To home in on local scale convective development, the Observatory is alsoconsidering the option of adapting a non-hydrostatic high-resolution model foroperational trial. Eventually, it is intended that all downstream applications, suchas dispersion models and windshear studies for aircraft operation, will all derivetheir products from one single parent model. As such, in addition to the regular

5

modelers, a pool of professional expertise from within the Observatory also needsto be assembled. As a first step, an open laboratory for forecast applicationsdevelopment is being set up so that a designated NWP interest group may worktogether for sharing of knowledge and experience. In time, they will offer technicalsupport to the modeling group and undertake NWP-related projects. On thehardware side, possibilities to run regional models on PC clusters will be explored.If feasible, this will provide a more flexible alternative to the more expensivesupercomputers.

2.2.2 Nowcasting System

For numerical prediction, the first two to three hours in the forecast periodwill always remain a blind spot because of the minimum data cut-off time and thefinite computation time required even on a high-speed computer. To bridge this gap,an operational nowcasting system SWIRLS (Short-range Warning of IntenseRainstorms in Localized Systems) is designed to assimilate observational data fromradar, satellite and raingauges to produce short-range rainfall prediction. The firstphase of SWIRLS, mostly radar-based applications, has been put into operationsince April 1999. When applied to the tropical cyclone cases in 1999, theprediction of rainfall within the next couple of hours is reasonably accurate in termsof location and amount (Figure 6). The cyclone circulation is also well representedby the analyzed wind field based on movement of radar echoes, thus providing anadditional tool for radar eye fix and monitoring of tropical cyclone movement.

In the future phases of SWIRLS, various options will be explored to extendthe rainfall forecast validity period. For echo movement, use of wind profilerinformation and model winds will be investigated. More sophisticated treatmentson rain intensity changes will be explored. For even longer shelf life, satelliteinformation ultimately has to be incorporated and the relevant preparatory work isalready under way. In conjunction with the development of a non-hydrostatichigh-resolution model, frequently updated local scale analysis combining NWP-derived first-guess fields with real-time data from the high-density observationalnetwork will also be explored. This will provide additional information forquantitative rainfall forecasts and operations of rainstorm warnings.

2.3 Operational Systems

Time has never been on forecasters' side. Given the sheer volume ofinformation available to the forecasters these days, time is, relatively speaking, ineven shorter supply. One can argue that if the information were not intelligentlypresented for ready digestion, then forecasters would find even less room for"thinking time" and any potential benefits of extra data would simply go to waste.

6

The Central Forecasting Office (CFO) in Hong Kong has chosen to solvethis problem using the rapidly developing web technology. Outputs from NWPmodels, data from automated observation networks, web cam images for visibilitymonitoring, Climatological data, satellite and radar images and their derivedproducts, etc., are centralized into an intranet and are easily viewed withcommercially available browsing software. The "homepage" concept, apart fromproviding a friendly and attractive users' interface (Figure 7), has actually broughtabout a more dynamic, flexible and interactive desktop working environment. Newgraphical tools can be readily introduced and implemented, ensuring forecasters'requirements can be met with minimal delay. Information is all set for retrieval anddisplay at the click of a button

Should an emergency disrupt the normal forecast operations, a back-upCFO purposely set up at another location is all ready to take over. Basic operationcan be maintained through Internet access to overseas data sources. All that isrequired is a couple of PCs equipped with forecast bulletin preparation anddissemination capabilities.

3. THE HONG KONG EXPERIENCE - SERVICE DEVELOPMENT

In Hong Kong, the emphasis in the past several years has been very much onthe development of a comprehensive public weather service. The objective is todeliver weather warnings and messages in a timely and efficient manner direct tothe users through the latest information technology. The strategy is to operatemultiple concurrent information paths to recipients in multimedia formats. Slowlybut gradually, a differentiated approach is emerging to meet different needs:unlimited access of information through automated enquiry systems, the massmedia or Internet for the public at large; and limited access of more technical detailsrestricted to special users through a combination of Intranet, dedicated servers andunlisted telephone hotlines.

3.1 Weather Warnings and Public Education

The Hong Kong Observatory operates a wide range of warnings onweather hazards, e.g. tropical cyclones, rainstorms, thunderstorms, landslip,flooding, strong monsoon, fire danger, cold snap and hot spell. A comprehensivepublic education programme has been developed through the years to increasepublic awareness. This includes information pamphlets, special press releases andarticles, educational material on the web and TV, organizing visits to theObservatory and setting up a resource centre for easy access to meteorological,climatological and other geophysical information.

7

3.1.1 The Mass Media

The mass media itself, of course, is not immune to the electronic revolution.The Observatory has developed coded warnings for transmission to variouscommunication media. Once captured, the appropriate weather warnings can beautomatically decoded, disseminated and relayed to the public in a timely andefficient manner. This transmission mode is particularly effective for theincreasingly popular mobile phones and pager networks, avoiding confusion thatmay result from ambiguous wordings in text messages, ensuring consistencyamong the various networks, and in a split second getting the important warningmessages literally to the hands of the end-users.

In an effort to reduce the social impact caused by inclement weather, earlyalerts of imminent changes in warning status and precautionary measures are alsoincluded in media broadcasts. For example, advice of health risks, navigationalsafety and social orderliness is spelt out in the Observatory's TV weather broadcastin hot or cold spells, reduced visibility and rainstorm situations respectively.

The black-and-white stereotyped newspapers weather columns in the olddays are becoming increasingly colourful and lively. To fit into the overall layoutdesign, many news editors have their own requirements of presentation style andcolour schemes. The Observatory has taken the initiative to understand their needsand preferences. Instead of one set of standard texts and maps for all, there is agrowing trend of tailor-made products of weather information and graphics beingspecially prepared and dispatched to the print media (Figure 8).

3.1.2 Conventional Channels

Apart from the traditional way of disseminating information through themedia, a dial-a-weather service provides current readings as well as latest warningsand forecasts. The service is a computerized operation with 200 telephone lines.Text-to-voice capability enables it to operate in three different languages anddialects, mostly in an automatic manner. About 25 million calls are handled eachyear. A separate interactive telephone answering system provides the public withready access to a diversity of information on weather, climatology, records, etc.,which they could choose on their own initiatives. A fax-on-demand service is alsoprovided. 3.1.3 Web-based Information Channels

Public weather services in the 1990s can no longer ignore the impact of theburgeoning Internet. Since its launch in 1996, popularity of the Observatory'shomepage (www.weather.gov.hk) continues to grow rapidly (Figure 9). Apart from

8

the standard fare of weather facts and forecasts, information can now be conveyedvisually and graphically in multimedia formats. Weather information products,such as temperature readings and rainfall maps, are made available in real time.

The Observatory's homepage has been voted the most popular Governmentweb site. During the last 12 months, a total of 38 million pages of information wereretrieved. The highest number of retrieval in a single day exceeded 2.1 million on16 September 1999 when Typhoon York hit Hong Kong.

TV weather programmes presented by Observatory meteorologists,previously very much confined to the cable TV network, now reaches out to a muchwider audience through a web version on the Observatory's homepage.

To enhance preparedness, information is regularly provided to governmentofficials for decision-making in weather-related situations. This has also taken theelectronic leap through the recent launch of a Government Weather InformationServer (GOWISE), again employing web technology. Information of appropriatetechnical content is provided for professionals and administrators. The homepageconcept also allows the easy provision of tailor-made information to suit individualgovernment departments (Figure 10).

At the airport, airline operators and flight crews can retrieve flightdocumentation and other information from a web-based MeteorologicalInformation Dissemination System (MIDS). Documents are retrieved by flightnumber, destination aerodrome, or flight route. They have also access to aerodromeobservations, forecasts and warnings, radar and satellite images, products from theWorld Area Forecast Centres, and tropical cyclones affecting the area.

MIDS is very popular among the users, who need only to use a dial-uptelephone line connected to a PC. On average, they download over 400 megabytesof data each day with a rising trend (Figure 11). It has recently gained therecognition of the International Civil Aviation Organization as a system enablingairlines to obtain products for their operation "with minimum investment inhardware, software and training".

Our experience in recent years is a testimony to the profound impact theInternet has on forecast operations. The number of direct contacts between theObservatory and its users is now about 470,000 per day or about 14 million permonth (Figure 12). Ten years ago, the Observatory typically handled only severalhundred inquiries per day mainly through the telephone. In many respects, theforecast centre in a meteorological service is no longer an information sourcestrictly for weather forecasts and warnings only. It now serves a much widerpurpose that includes other weather-related facts and stories. Through the Internet,

9

an interactive relationship emerges between the forecasters and the general public.Forecast and warning messages can be conveniently and directly transmitted tohomes and offices; in return, the public expects more real time information onvarious weather aspects to be made available on their computer monitors for theiruse and reference. At the Observatory website, for example, the number of hits onweather readings is actually higher than the forecast itself. Through these trends,we can keep track of the public's expectation and demand; and to maintain anefficient service, our product availability and delivery should be able to meet suchneeds.

3.2 Publicity and Media Initiatives

Against a growing list of competitors, ranging from well-established privateconsultants to the weather enthusiast next door, it is very important for an officiallydesignated meteorological service to uphold its authority and authenticity. Ifhigh-profile forecasts such as those in severe weather situations were correct, theywould obviously help to boost the service's popularity stake. But despite the best ofintention, we cannot be right all the time. If the credibility of the service hangs onlyon severe weather forecasts, then it would be very much a case of hit or miss. Whatwith severe weather not likely to occur for more than a few times each year, theopportunities to enhance the profile of the service therefore really have more to dowith how we interact with our customers during the prolonged "peace time" periodsthat cover most part of the year.

At the Observatory, we recognize the importance of making "fair weatherfriends" when the pressure is off, taking a proactive approach in establishingworking relationship and mutual trust with the public, the media and other users.

For the public, we routinely monitor and respond to letters and commentsin the newspapers, phone-in radio programmes and weather forum on the Internet.A special liaison group named "Friends of the Observatory" is now in its fourthyear of operation, with membership exceeding 3,400. Regular activities, mostlytalks, visits and newsletters, are organized for the members. Members are alsoinvited to take an active role such as serving on selection panels for best TVweather presenters, providing feedback on our services, and contributing articles tothe newsletter.

For the media, apart from interviews on topical issues, we also have aplanned series of press information on various subjects through the seasonal cycle.For better mutual interaction, a "one-stop shop" is set up - an officer from theObservatory is designated as a focal point for liaison. An added benefit from such avibrant relationship is that the media's professional help can also be enlisted inpromoting the work of the Observatory when the opportunity arises.

10

4. COOPERATION TOWARDS A BETTER FUTURE

In Hong Kong, we are appreciative that significant development inforecasting techniques and public weather services is only possible as a result of thefree and unrestricted exchange of data. For the protection of life and propertyagainst weather impact, the principle of free and unrestricted exchange of data mustbe preserved. This is a necessary condition to bridge the technological gap amongthe NMHSs and to encourage the science and services to flourish.

Looking into the future, information technology will enable the public tohave ready access to a variety of information sources. But ready access may alsocome at a price. The public at large could well be inundated with all kinds ofinformation - truths, half-truths as well as untruths. The likely scenario in the nextmillennium is for the man in the street to ask "Which weather forecast should Ibelieve in?" rather than "What is the latest weather forecast?".

It is important that established meteorological services should maintain ahigh profile so that the man in the street will always know where to turn to forauthentic information. In a climate of privatization and commercialization, there isa tendency to under-estimate the intangible value of an officially designatedmeteorological service with strong links to the government administration. Anofficial identity serves to project an aura of authority, reliability and accountability,thereby ensuring concerted action in emergency preparedness and response. Thedanger is for designated meteorological services to retreat into a corner, leavinggrounds for others to exploit and at the same time losing the public's trust andrespect cultivated over a long period of time.

On the positive side, if we take full advantage of our well-establishedlogistic and technical infrastructure, and as long as we remain mindful of theconstant need to improve operational and service standards, NMHSs will, andshould, continue to play a leading role in the millennium to come. And if suchservices can collectively pool their resources together, streamlining on areas ofredundancy and striking out strategically at new areas of development, then thepossibilities are even more exciting.

4.1 Technology Exchange

While the cost of telecommunication and computing is coming down toaffordable levels, the cost of data collection is on the rise. As a result, manyNMHSs have found it difficult to upkeep their observational network andequipment. We have to look for cheaper alternatives without sacrificing accuracy

11

and reliability. In Hong Kong, our solution for surface data is a low-cost custom-built network of automatic observing stations. As part of the WMO VoluntaryCooperation Programme (VCP), the Observatory would be happy to share ourexperience in the design and operation of such automated observing systems.These systems will dispense with the expensive deployment of trainedprofessionals to outstations. They are very much left alone apart from occasionalmaintenance visits or, if so desired, can also be designed and set up for simpleoperation by personnel with no meteorological training.

4.2 Access to Operational Meteorological Data

Recent advances in web-based technology provide an opportunity to makean immediate and affordable quantum jump to bridge the technological gap amongthe NMHSs. The web is a low-cost and effective way to exchange information,from data to graphics to software. It will probably become the major means for theexchange of meteorological information in the coming years, even for data orwarnings on a real time basis. Although how far such technology can effectivelydeliver information to the general public will depend on the networkinginfrastructure within individual countries, there should be at least an electroniclink-up among the NMHSs. More password-controlled data servers should also beset up at world and regional centres as well as other NMHSs from which ready-to-use actual or prognostic information, as well as a variety of forecasting tools andsoftware, can be retrieved via the internet. This engineering approach will buy timefor on-going capacity building programmes to bear fruit.

More convenient and affordable telecommunication possibilities also meanthat NMHSs can reap mutual benefits through the sharing of locally available data(such as Doppler radar, lidar, wind profiler) at a higher temporal resolution.Unconventional data and information from other sources can also be exploited.One example is AMDAR (Aircraft Meteorological Data Relay) observations in theaviation sector. The AMDAR programme has been implemented in a number ofcountries in RA I, IV, V and VI. It is a rich and potentially affordable source ofinformation for ingestion into numerical modeling systems to improve datacoverage over oceanic regions.

In Hong Kong, working with local airlines and aviation communicationoperators, the Observatory is studying the feasibility of setting up one suchprogramme to collect automated upper-air meteorological observations from civilaircraft flying over the South China Sea. If the programme is implemented,information gathered can be disseminated on GTS. Attempts will be made to ingestthe data into the Observatory modeling system. Wind and turbulence data obtainedin the vicinity of the Hong Kong International Airport will also be up-linked tolanding and departing aircraft.

12

On the public services side in an increasingly globalized and mobileenvironment, either through physical travelling or electronic net-surfing, readyaccess to reliably reported information will allow individual NMHSs to compilebulletins such as world cities temperatures and forecasts for the consumption ofusers within their own community. This will encourage the use of official forecastsand enhance the integrity and profile of NMHSs.

4.3 Pooling of Resources

Another possibility to pool resources of NMHSs for streamlining areas ofredundancy and hitting out strategically at new areas of development is innumerical weather prediction. From a selection of open community numericalmodels, with different plug-in modules and package suites for customization, it isbecoming increasingly viable for individual NMHS to run purposely built limitedarea forecast model on high performance PCs or PC clusters. However, for dataassimilation and initialization, we need processed self-consistent gridded datafields that have been quality-checked. We can choose to do this within our ownNMHS, of course. But this inevitably will require more computing resources andalso the necessary expertise. And even then, we will only have part of the pictureowing to data availability problems and constraints. If a regional data analysiscentre(s) can be established to collate and analyze the data, from conventional toremote sensing sources, and to provide quality output products in real time or nearreal time for operational use, then the application of NWP techniques will no longerbe the impossible dream for many, and more services will be able to run some sortof regional mesoscale models on moderate means.

Entitlement to the use of the analyzed products should be on the basis ofraw data contribution. And the following advantages would immediately becomeobvious:

(a) NMHSs will have the motivation to maintain a good-quality monitoring andobservational network, and will be in a strong position to secure continuingsupport and funding from their respective governments;

(b) from the science angle, the pooling and combined analysis of data will bringabout a more complete input picture and hence more quality outputinformation can be expected;

(c) the application of input analyzed data from a common source will facilitatethe inter-comparison of algorithms and models, providing useful indicatorsfor further tuning and refinement of forecast packages;

13

(d) the integrity and profile of NHMSs can be enhanced against competitorsfrom private and international sectors.

We live in an age of great changes. The arrival of the millennium is a

less-than-subtle reminder that we are, like it or not, stepping forward into a new era.We have successfully navigated around the Y2K-compliance issue through thefiddling of numbers. But as we surf into the new age, the well being of NMHSsrequires something that is more durable and purposeful. And we need to have deepresolve and thoughtful planning on our part in order to meet the challenge.

14

Figure 1 Hong Kong - a small city along the coast of southern China.

15

Figure 3 3-D radar image of Typhoon York (9915) with eye over Hong Kong at10:30 a.m. on 16 September 1999.

Figure 2 Dual-Doppler analysis of the wind field of Typhoon Maggie (9906)during its passage over Hong Kong at 3:36 a.m. on 7 June 1999.

16

Figure 4 Network of weather sensors for the Hong Kong International Airport.

17

Figure 5 RSM-derived products: forecast time series at grid point near Hong Kong(upper panel) and automated worded forecasts and warnings (lowerpanel).

18

Figure 6 1-hour rainfall forecast by SWIRLS for Typhoon Maggie issued at 3:30HKT on 7 June 1999 (upper panel) and the corresponding one hourrainfall accumulation reported by the raingauges from 3:30 HKT to 4:30HKT on 7 June 1999 (lower panel).

19

Figure 7 Synop Coding Panel on Web-based Forecast Operations System (upperpanel) and Intranet Index Page (lower panel)

20

Figure 8 Sample product prepared for newspaper

21

Figure 9 Significant increase in information retrieval through the Internet at HongKong Observatory's web page.

22

Figure 10 Sample of the front page of the Government Weather Information Server

23

Figure 11 Trend of web-based retrieval of flight document and aviation weatherinformation.

24

Figure 12 Trend of the total number of direct contacts to the Hong KongObservatory services.