from observations to forecasts – part 7. a new meteorological monitoring system for the united...
TRANSCRIPT
272
Wea
ther
– O
ctob
er 2
010,
Vol
. 65,
No.
10
From Observations to Forecasts – Part 7. A new meteorological monitoring system for the United Kingdom’s Met OfficeAidan GreenMet Office, Exeter
The United Kingdom’s Met Office owns and operates a network of over 200 auto-matic weather stations (AWSs). In 2003, a project was initiated to deliver a modern, value-for-money and sustainable replace-ment surface observing system, to be known as the Meteorological Moni toring System (MMS). Key benefits of MMS include the availability of one-minute time-resolution data, increased monitor-ing and control from a central system, and greater overall flexibility to add new sites and sensors. The project considered vari-ous options, including continuing with the existing multi-system network, design-ing a new system in-house, or purchasing a new system from a third party supplier. The third option was selected, and in 2005 an advert was placed in the Official Journal of the European Communities (OJEC) inviting expressions of interest in supplying a new AWS solution to meet the Met Office’s requirements. This article provides a brief background to the evolu-tion of the Met Office’s automatic surface observing network and a description of its new MMS.
The users and requirements for surface meteorological observations are exten-sive and varied. Observations are of fun-damental importance to the work of the Met Office, and are essential to the fore-casting process since they are required to initialise weather and climate predic-tion models. Observations are also used to verify the performance of these mod-els and to provide a ‘ground truth’ to vali-date satellite observations. They are also used by forecasters in near real-time, helping increase the accuracy of fore-casts for the next few hours – a process known as ‘nowcasting’. However, the users of observations include not only forecasters and climate scientists, but academic, defence, health, media and transport communities, as well as the general population and the many com-mercial organisations impacted by weather and climate.
Background to Met Office automatic surface observ -ations in the UKIn the 1970s, the Met Office operated a syn-optic surface observing network consisting of approximately 100 stations. Observations were transmitted hourly, and all observa-tions were manual. The network distribution was uneven with large data-sparse areas in Scotland, the English Lake District, Wales and Northern Ireland (Figure 1(a)).
A surface observing station providing near real-time data to forecasters helping them form a synopsis of the current weather situation is known as a ‘synoptic’ station. Stations providing delayed data are useful for climate purposes, but not for synoptic purposes. Such stations are known as ‘cli-mate’ stations. Data from synoptic stations can be used for climate purposes, depen-dent on their quality, but not vice versa. Figure 1(a) is a map of the Met Office’s syn-optic network on 1 January 1970. For com-parison, Figure 1(b) is a map of the Met Office’s synoptic network on 1 January 2010. The maps clearly illustrate the expansion of the synoptic network over this period, which has only been possible due to the introduction and development of AWS.
In the 1980s, a microprocessor-based measurement, coding and transmission sys-tem, called the Synoptic Automatic Weather Station (SAWS), was introduced. SAWS auto-matically measured wind speed and direc-tion, dry-bulb and wet-bulb temperature, barometric pressure and rainfall amount. During the 1980s over 40 SAWS were deployed within the identified data-sparse regions, creating a more evenly distributed synoptic network. The SAWS system was later updated to enable the measurement of cloud height, visibility and global solar radiation, and became known as the Enhanced Synoptic Automatic Weather Station (ESAWS; Figure 2).
The Met Office also operates a network of Severe Icing Environment Synoptic Automatic Weather Stations (SIESAWS) at a small number of UK sites located in extremely remote mountainous terrain, such as on the summit of Cairngorm in the
Figure 1(a). Met Office synoptic network as on 1 January 1970.
Figure 1(b). Met Office synoptic network as on 1 January 2010. Stations marked with red dots are those stations that existed in 1970 and are still operational in 2010.
273
Weather – October 2010, Vol. 65, No. 10
New AW
S for UK Met Office
Scottish Highlands (Figure 3), where hourly data from a limited set of sensors (air tem-perature, humidity and wind speed) are logged and reported.
The 1970s, 1980s and 1990s were periods of considerable technological change which saw the introduction and retirement of sev-eral different automatic observing systems. These included the Digital Anemograph Logging Equipment (DALE), a five-channel data-logging system which automatically processed and recorded wind data to mag-netic tape. DALE was superseded by a more modern microprocessor-based wind-only reporting system, the Wind System Observer Processor (WSOP), which remained in use at several sites until the introduction of MMS. Another automatic system, which was designed to measure more parameters than wind alone, was the Agromet Climate Recording Equipment (ACRE). ACRE was replaced in the early 1990s by the Limited Capability Climate Recorder (LCCR), which itself was replaced by a system known simply as the Climate Data Logger (CDL: Figure 4). The Met Office established a net-work of over 40 CDLs, a system based on a commercial off-the-shelf logger, manufac-tured by Campbell Scientific Ltd, which recorded a range of basic meteorological elements including air temperature, relative humidity, rainfall, radiation, wind speed and direction, and surface and soil tempera-tures. Although the CDL was originally designed for use at climate observing sta-tions, the network expanded to satisfy two further purposes: filling gaps in the synoptic surface observing network, and meeting commercial customers’ location-specific requirements for meteorological data. A personal computer (PC) located at Met Office headquarters would automatically contact the sites via modems and collect data on an hourly basis. The computer would then centrally encode and transmit the data in the World Meteorological Organization (WMO) defined alpha-numeric SYNOP format (known by WMO as FM-12) designed for the exchange of meteorologi-cal measurements made at automatic and manned weather stations (WMO, 2009).
In the 1990s, a PC-based AWS was designed in-house by the Met Office as an aid to manual observation, helping the human observer combine manual and auto-matically observed elements to produce a single hourly quality-controlled report. This Met Office AWS system – the Semi-Automatic Meteorological Observing System (SAMOS) – initially only supported the same range of sensors as the ESAWS system. A series of distributed interface modules connected the senors to a local area network (LAN), through which the PC would collect the data before processing, storing it within a local database and transmitting to Met Office headquarters. The SAMOS (Jones
Figure 3. Cairngorm Summit SIESAWS. Sensors include a sonic anemometer (top centre of image), a humidity sensor (within tube at top left of mast – the tube does not ice up as easily as the temperature screen), and four temperature sensors: two within a screen (iced up and not visble in this image) and two on ‘whip’ aerials which vibrate in the wind and so help to prevent build up of ice (one of the whip aerials is visible at the top of image to the left of the anemometer). (© Crown Copyright, the Met Office.)
Figure 2. Hawarden Airport (Chester) ESAWS. Sensors visible include, from left to right, a yellow laser cloud base recorder, a white visibility sensor, the Stevenson Screen containing wet- and dry-bulb thermometers, green cabinets containing data logging equipment and a pressure sensor, static pressure head system (three-metre tripod with white wind vane on top), green tipping-bucket rain gauge and climbable wind mast with wind sensors at top. (© Crown Copyright, the Met Office.)
274
Wea
ther
– O
ctob
er 2
010,
Vol
. 65,
No.
10
New
AW
S fo
r UK
Met
Off
ice
et al., 1994) evolved throughout the 1990s to support fully automated operation and to meet a number of special requirements, including ten-minute reporting frequency
at some sites, supporting modern sensors such as visibility sensors (also known as visi-ometers), present-weather sensors and snow-depth sensors, as well as new software
developments, such as automatic cloud cover amount algorithms. A key difference in architecture to the CDL was that SAMOS processed and converted its measurements into WMO FM-12 format on-site (rather than at Met Office headquarters).
SAMOS (Figure 5) was installed at over 100 stations across the UK, at a mixture of fully-manned, part-manned and fully auto-matic sites. The SAMOS display (Figure 6) did not display graphical wind data, so yet another separately developed in-house wind display was required, a system known as the Mark 6 wind display (Figure 7).
Throughout the early 1990s, a number of sensor additions were made to the ESAWS and SAMOS systems, including soil, grass and concrete temperature sensors and lim-ited deployment of solar radiation sensors (also known as pyranometers). In the latter half of the 1990s, visibility sensors and laser cloud base recorders (LCBRs, also known as ceilometers) were introduced into the net-work, and were deployed at almost all ESAWS and SAMOS hourly synoptic sta-tions. Techniques for estimating total cloud cover amount using clustering and averag-ing techniques were developed, and these algorithms were deployed at SAMOS sta-tions fitted with LCBRs. Since 2000, present-weather sensors, snow-depth sensors and next-generation LCBRs have also been added to many SAMOS stations. In 2001, the SAMOS software was further developed to modify present-weather sensor output as necessary based on evidence from the other SAMOS sensors, a functionality known as the ‘present-weather arbiter’ (Lyth, 2006).
In summary, at the time of deciding to introduce the new MMS, the Met Office sur-face observing network consisted of over 200 sites, of which approximately half were SAMOS, one-quarter were ESAWS and another quarter CDLs, complemented by a small number of SIESAWS and WSOP sites. The separate Mark 6 wind display system was also required at manned sites. There were LCBRs and visiometers at around 70% of stations, automatic present-weather capability at 25% of stations, sunshine and radiation sensors at around 40% of sites, and snow-depth sensors at 20%. The net-work also benefited from supplementary voluntary and auxilliary observations recorded at additional locations (such as coastguard stations and health resorts) to help satisfy site-specific forecasting require-ments, climate network density require-ments and other business needs.
Decision to procure a new AWS systemUntil very recently, the Met Office surface observing network was based on a combi-nation of various systems, relying on
Figure 4. East Malling (Kent) CDL. Sensors visible include, from left to right, a green tipping-bucket rain gauge, the climbable wind mast with wind sensors at top, the Stevenson Screen containing dry-bulb thermometers and humidity sensor (with data logger attached to the bottom of the screen stand) and white stakes marking various surface and sub-surface temperature measure-ments. (© Crown Copyright, the Met Office.)
Figure 5. Andrewsfield (Essex) SAMOS. Sensors visible include, from left to right, a green present-weather sensor (twin-headed unit with precipitation sensor mounted on top), lowerable wind mast with wind sensors at top, Stevenson Screen containing thermometers and humidity sensor, white stakes marking various surface and sub-surface temperature measurements, green tipping-bucket rain gauge, snow-depth sensor (downward pointing silver coloured acoustic echo sensor attached to small vertical pole), yellow laser cloud base recorder and white visibility sensor (twin-headed unit). (© Crown Copyright, the Met Office.)
275
Weather – October 2010, Vol. 65, No. 10
New AW
S for UK Met Office
increasingly diverse and obsolescent tech-nology. The systems were ageing: SAMOS has been operational for 15 years and ESAWS for over 20. As a result the network was gradually becoming more difficult and costly to enhance, manage and maintain. The previous systems relied heavily on local on-site data processing, leading to major expense when introducing upgrades. Although the instruments and certain parts of the processing software were able to cope with continuous data flows, they were fundamentally linked to hourly reporting and transmission of data using old alpha-numeric coding conventions (such as FM-12) which have been in existence since observations were produced manually with-out any automatic aids. This was not ideal given the growing demands of nowcasting
systems and developments in numerical weather prediction (NWP) model data assimilation, which require more frequent and flexible reporting of observations and new data types, such as precipitation drop-size distribution, not easily encoded within FM-12.
ESAWS in particular was no longer a flex-ible system. It was extremely difficult to make changes; programming updates were not possible, and various modern sensors (present-weather sensors, for example) could not be added. Another problem was that it was becoming increasingly difficult to source spare parts for maintaining these ageing systems. To continue with the exist-ing networks would have resulted in a sys-tem ever more reliant on costly maintenance resources, and a network which would have
increasingly failed to meet growing user expectations.
In order to address the varying needs of the many users of land-based observations, the Met Office required a more versatile and scalable observing infrastructure – one where expansion or contraction of the net-work and/or the range of sensors at a given site incurred minimal marginal costs. It was envisaged that this would be best achieved by taking full advantage of electronic sens-ing (by getting the best value from sensors which provide virtually continuous mea-surements) and by carrying out far more data processing centrally.
A project was initiated to consider the various options, including continuing with the existing network, designing a new system in-house - or purchasing a new sys-tem from a third-party supplier, which proved to be the preferred way forward. This option had the benefit of combining the data telemetry and Supervisory Control and Data Acquisition (SCADA) expertise of the supplier, the data-logging expertise of its sub-contractor and the meteorological expertise of the Met Office.
The Meteorological Monitoring System (MMS)The Meteorological Monitoring System (MMS) was supplied by CSE Servelec and is based on SCOPE-X, its well-established data telemetry and SCADA suite of soft-ware. SCOPE-X consists of a resilient real-time database, connected to an Oracle long-term historic database, and features flexible data interfaces. SCOPE-X is cur-rently widely used within the water and utility sectors for other wide area applica-tions, with many other current UK users including the Environment Agency, National Grid Gas, Severn Trent Water and Trinity House.
The core of the solution comprises two sets of central server computers acting as a duty/standby pair located in primary and back-up computer halls. This arrangement ensures that if one server or computer hall has technical problems, the system will con-tinue to operate seamlessly and data will remain available to downstream systems and users. MMS also contains a pair of data-base servers, providing a 12-month rolling archive of one-minute resolution data from all sensors at all sites, and web servers which enable desktop users to access information within the real time and historical databases.
The central servers provide extensive data monitoring, automatic alarms, calculation of derived values from embedded and basic user-definable algorithms, and various facil-ities for implementing centralised automatic control. Statistical functions and conse-quential alarm generation are also standard
Figure 6. SAMOS latest observations display. (© Crown Copyright, the Met Office.)
Figure 7. The Mark 6 wind display. (© Crown Copyright, the Met Office.)
276
Wea
ther
– O
ctob
er 2
010,
Vol
. 65,
No.
10
New
AW
S fo
r UK
Met
Off
ice
features of the SCOPE-X system. Alarms may be generated from any values within the system, including raw data, derived data and statistical results.
The system caters for configurable sched-uling and prioritisation of data-logger communications, and the automatic polling of loggers utilises existing Met Office com-munications infrastructure, a mixture of broadband, mobile and standard fixed landlines.
Campbell Scientific sub-contracts to CSE Servelec in supplying the MMS, and has pro-vided the CR3000, CR1000 and CR800 data loggers for installation at AWS sites. Campbell Scientific has over 30 years of meteorologi-cal experience, and proven reliability in the field. Its new generation of data-acquisition units offers a powerful and flexible pro-gramming interface which enables them to be readily interfaced with the SCOPE-X cen-tral system. Strangeways (2004) provides a
Figure 8. MMS latest observations display. (© Crown Copyright, the Met Office.)
Figure 9. MMS wind display. (© Crown Copyright, the Met Office.)
history of the use of data loggers for meteo-rological data collection and an insight into their operation. The loggers interface with all existing sensors at sites and are flexible enough to accomodate the addition of new sensors in the future. The loggers collect, process and store data from all sensors, which are then retrieved by the central serv-ers in Exeter where they are further pro-cessed centrally to generate and transmit the hourly FM-12 coded messsages. A major difference between MMS and previous sys-tems is that MMS retains the high-resolution data which are of considerable value to vari-ous users (e.g. forecasters, research scien-tists, engineers and quality control staff ). Data sampling is actually carried within the loggers at a variety of sub-minute frequen-cies in line with WMO recommendations, before storing one-minute averages along with associated quality indicator flags within the data archive. Further numerous quality control checks are applied within the central system prior to producing and transmitting the FM-12 coded messages.
MMS also retains the ability, pioneered in SAMOS, for observers to manually enter supplementary data (such as cloud type) via a PC and to apply quality control to auto-matic observations. The user workstations are high-resolution colour PCs, running a web-browser-based user interface within a standard Windows operating environment. The application utilises all of the standard tools within a web-browser (such as back-ward, forward, printing, favourites and mul-tiple windows) while still retaining the speed and stability of a standalone applica-tion. A uniform appearance and operation is provided throughout all windows (Figures 8 and 9), providing a consistent means of access, interface and operation, and an easy-to-use interface for all MMS users. The MMS system also includes several graphical wind display options (Figure 9), something that was only possible with a separate system previously.
In accordance with the GCOS Climate Monitoring Principles (WMO, 2003), the impact of the introduction of this new observing system was carefully assessed prior to implementation, including side-by-side trials with previous systems. Detailed analysis for over a year of side-by-side run-ning of MMS at Pershore College (Worcs) CDL (a Central England Temperature site) and at the Camborne (Cornwall) Met Office SAMOS site showed that there was an excel-lent level of agreement between the MMS and SAMOS or CDL systems (Clark et al., 2010). Comparison of all measured parameters (in both one-minute data and in coded mes-sages) between MMS and both of these systems showed very close agreement, with the magnitude of any differences being well within the target accuracy of the observations. Laboratory bench tests were
277
Weather – October 2010, Vol. 65, No. 10
New AW
S for UK Met Office
also conducted to compare the wind meas-urements of the MMS, SAMOS, CDL and ESAWS systems. This work concluded that the upgrade to MMS represents a huge improvement in the accuracy of the calcula-tion and storage of average wind speed (Sloan, 2010). Publication of detailed results from these inter comparisons is planned in the near future. Metadata, describing where and when the upgrades to MMS have taken place, have also been recorded as part of the permanent metadata record for all sites.
SummaryMMS has replaced the previous diverse data logging systems with a new single-system network, with a more uniform set of sensors and data-processing algorithms. The new loggers interface with all existing sensors across the network and provide increased flexibility to adapt to new sen-sors in the future. Manned sites continue to be provided with the functionality to display and interact with automated mea-surements. The new SCADA software is located within a highly resilient central sys-tem at the Met Office headquarters in Exeter. This central system enables data col-lection, monitoring and control at frequen-cies capable of being adapted to meet changing operational requirements. The system is designed to be easily scalable, allowing the number of sensors at sites, and the number of sites themselves, to be increased or reduced as necessary. The sys-tem has been carefully designed (learning
from the strengths and weaknesses of pre-vious systems) such that its capacity can be easily expanded to handle increasing amounts of data and to interface with downstream Met Office data storage sys-tems. It has also been designed with mul-tiple data storage buffers to ensure completeness of record should any indi-vidual components fail.
Procuring and implementing a new AWS system has proved to be a large and com-plex undertaking; however, the benefits to the Met Office are already clear to see within the system’s performance measures. The new system has replaced several ageing and diverse observing networks with a modern, unified, system which provides greater net-work scalability, greater flexibility of sensor deployment, the collection of higher resolu-tion data and vastly improved centralised end-to-end system monitoring and control.
AcknowledgementsThanks are due to various Met Office col-leagues, including Paul Clews and Luke Green for providing the maps, and to all of those who provided comments on drafts of the article, notably Colin Chatters, Stuart Goldstraw and Mike Molyneux. Correspondence to: Aidan Green,
Met Office, FitzRoy Road, Exeter, UK, EX1 3PB
© Crown Copyright, 2010, published with the permission of the Controller of HMSO and the Queen’s Printer for Scotland
DOI: 10.1002/wea.649
ReferencesClark M, Legg T, Evans L, Pethica C, Lee D. 2010. MMS, SAMOS and CDL Comparisons: Final Report. Met Office Internal Report. 22 March 2010.
Jones DW, Wright AM, Whiten BA. 1994. SAMOS. A state of the art observing system. WMO Instruments and Observing Methods, Report No. 57, WMO/TD-No.588. pp. 63–67.
Lyth DR. 2006. Results of using Present Weather instruments in the United Kingdom. http://www.wmo.int/pages/prog/www/IMOP/publica tions/IOM-94-TECO2006/P1(19)_Lyth_UK.pps [Accessed 28 June 2010].
Sloan C. 2010. Uncertainty of wind speed measurements in Met Office observing systems: Lab-based testing. Met Office Internal Report. 5 March 2010.
Strangeways I. 2004. Back to basics: The ‘met. enclosure’: Part 10 – Data loggers. Weather 59: 185–189.
WMO. 2003. GCOS Climate Monitoring Principles. http://www.wmo.int/pages/prog/gcos/documents/GCOS_Climate_Monitoring_Principles.pdf [Accessed 28 June 2010].
WMO. 2009. Manual on Codes, Volume I.1 Part A – Alphanumeric codes. http://www.wmo.int/pages/prog/www/WMOCodes.html
Frank Le BlancqJersey Meteorological Department
To make reliable projections of the future state of our climate, we need a better understanding of the past, so using knowl-edge of past climates to qualify the nature of ongoing changes has become a concern of growing importance during the last decades (Le Treut et al., 2007, page 102). By reveal-ing the factors that led to natural variation of climate in the past, we can more fully explain current and future climate trends. A considerable amount of meteorologi-cal data is digitised, but much of it covers only the last 50 years or so. Digitised data for years prior to about 1960 tends to be restricted to a smaller selection of stations with long series of records. Prior to 1850,
comparatively few instrumental records exist, yet in some countries a great deal of meteorological information awaits use or discovery.
In recent years, attention has been drawn to the wealth of regular observations con-tained in naval logbooks for instance, often over wide areas of ocean where data are oth-erwise very sparse or non-existent. Wheeler and Suarez-Dominguez (2006) were able to identify climate trends from observations in logbooks from as early as 1685 to 1700. In another pilot study, Brohan et al. (2009, page 227) demonstrated that using obser-vations from Royal Navy logbooks as late as 1938 to 1949 has substantially improved the observational coverage for the Second World War. Aside from this potentially rich maritime source, much additional terrestrial information remains languishing in diaries,
ledgers and observation books or as paper manuscripts. Scientists, as well as military and medical personnel, compiled some of the early records, while others were penned by interested and often well-informed pri-vate individuals. However, many run the risk of deteriorating beyond the point of use or simply being lost. This is especially so in less developed countries, where funds are generally very scarce, but can be the case elsewhere, because digitising histori-cal data is usually given a low priority and its importance not always recognised. This short paper describes the recent rescue and digitisation of Jersey surface pressure data and some potential uses.
The data rescue was undertaken in con-nection with the Atmospheric Circulation Reconstructions over the Earth (ACRE) sur-face data initiative following an enquiry
Rescuing old meteorological data