NATION WIDE METEOROLOGICAL NETWORK Foeke Kuik (1) Clarisa Martínez (2) (1) Telvent Netherlands, Landzichtweg 70, 4105 DP Culemborg, The Netherlands, foeke.kuik@telvent.abengoa.com, (2) Telvent Energía y Medio Ambiente, S.A. Dep. Meteorología, Valgrande 6, 28108 Madrid, (Spain), clarisa.martinez@telvent.abengoa.com Abstract Classic meteorological networks are usually based on manned and automatic weather stations, that send their meteorological reports to a National Meteorological Institute (NMI). Some important limitations of such networks were the low time resolution of the reports, the limited amount of data available for users, and the low level of integration with other systems that became available. New modern techniques, further automation and more integration will chance the future meteorological networks into Meteorological Observations Information Systems (MOIS). In this paper the new generation of meteorological networks (MOIS) is discussed. Some examples showing that these developments are already reality are shown, based on Telvent s MetConsole solution implemented at the Dutch Met Office KNMI and MeteoSwiss. 1. Introduction Synoptic meteorological networks used to consist of a number of weather stations, manned and unmanned. The weather stations contained a group of sensors and a logger that acquired the data from the sensors at regular intervals. The logger was an intelligent device capable of making a SYNOP (and sometimes other reports as well). If the station was unmanned, the SYNOP was forwarded to the NMI for example via telephone lines (PSTN), satellite communication for remote stations, or via some other means of communication. If the station was manned, an observer would have some device for displaying the data from the sensors, the observer would add his observations (clouds, visibility and weather), make the report(s) and forward it to the NMI. There are clear disadvantages in this approach. Firstly, SYNOPs contain low time resolution data, whereas the stations contain much more measured data. SYNOPs only contain data averaged over a certain period of time. Secondly, the data forwarded to the NMI is not real time data, but it is always old. Data of short term weather phenomena are not available at NMI, even though this data can be extremely important. Thirdly, from a technical point, it is inconvenient that every weather station produces its own reports. These reports are not identical for all stations, and if the WMO introduces changes, it takes a lot of effort to upgrade all weather stations. 2. Goals for new networks When a NMI plans to install a new meteorological network, the goals for the new network should be identified first. Just replacing the old systems without looking at the possibilities of new modern techniques and new sensor is a lost opportunity. Some of the goals could be: 1. Centralize report generation (easier maintenance of report definitions, simpler loggers, data transfer from the stations instead of SYNOPs). 2. Build a system that operates based on meteorological data instead of reports (users can have the availability of high time resolution data in semi-real time; reports become secondary products to be submitted to the GTS);
3. Ingest data from other sources so that information can be combined (lightning detection systems, weather radar, radio sondes, satellite images, etc.; this can be used to, for example, make strongly improved WaWa codes); 4. Ingest data from other networks to homogenize report production (for example, hydrological stations for which SYNOPS are produced, road weather information systems RWIS, etc.); 5. Combine other types of data, for example, data from a network of video stations, or cameras at weather stations. 6. Automate systems at regional airports (AWOS, AUTO-METAR, AUTO-TREND). 7. Automate all SYNOP stations by adding smart instruments like present weather sensors, ceilometers, etc). 8. Implement semi real-time data validation in the centralized systems that contain all the data, combining the data from various sources as well. 9. Integrate maintenance in central system (central monitoring of all stations, centralized maintenance, software and configuration upgrades from one central location). 10. Integrate airport aviation systems as part of the network (AWOS for airports can be setup in an identical way as a nation wide AWS network system). 11. Build the new system based on well accepted standards such as Ethernet, LAN/WAN technology and TCP/IP. Off the shelf computer systems can be used, reducing the costs, and all well know advantages of these techniques are then available (real time data exchange, network operations for monitoring and maintenance, well know user interface). This is not a complete list. Issues such as availability of certain types of data communications (leased lines, PSTN, ISDN, GSM, GPRS, UMTS, satellite,..) are not mentioned, but have to be considered from a technical and cost point of view. Another issue is the tendency to automate as much as possible, creating tension between users that have the opinion that only observers can provide certain data properly and instruments are not capable of this. Individual NMIs will have to consider these issues for themselves and include the outcome in the final system design. In Figure 1 below, a schematic picture is shown of a nation wide MOIS with all its components. 3. Components of the MOIS 3.1 MetConsole As mentioned in the abstract, a system as described already exists in the Netherlands and Switzerland and it is called MetConsole. MetConsole consist of the following components. 1. MetConsole Servers MetConsole Servers run the MetConsole Server software. This is Windows bases software with the capability to do data acquisition and processing, it contains a module for AWOS (airport meteorological systems), for ATIS (aerodrome Terminal Information System), and LLWAS (Low Level Windshear Alert System) systems. In general MetConsole servers acquire data from all data supplying systems, stores this data, and processes all data to obtain derived products such as reports, statistics, etc. 2. MetConsole Clients MetConsole Clients have access to all data on the MetConsole Servers. The software for servers and clients is identical, the different functionality for the servers and clients is implemented via the system configuration database. Clients are used for observers, forecasters (real time access to nation wide meteorological data), maintenance staff, and system administrators (ATIS and LLWAS: also for ATC staff). 3. AWS The Telvent made Automated Weather Stations are part of the system. Specific features have been implemented (for example alarm generation for system monitoring, also meteorological alarms) that work best with MetConsole software. However, MetConsole servers can work with any type or brand AWS.
Figure 1 A fully integrated nation wide Meteorological Observation Information System. The Data Acquisition Servers at the NMI collect the data of all data supplying systems. The software in the servers processes all data to generate reports, they perform data validation, send relevant data for airports to the airports servers, etc. The NMI Regional Centers do the same, but for certain regions in a country. All relevant data from the RNMIs are also forwarded to the NMI Headquarters. Airports have the same architecture with servers and clients as the NMI Headquarters and the NMI Regional centers. The software for all server and clients is identical, also for the LLWAS (windshear warning system) and ATIS (Aerodrome Terminal Information System) systems. Systems like weather radars, radio sonde systems, lightning detection systems, etc., usually operate their own acquisition computers, which in turn, provide their data to the MOIS servers. 3.2 NMI Headquarters The MetConsole software contains modules for Data Acquisition and Processing for Automated Weather Stations, including report generation (SYNOP, METAR, local reports); Data acquisition of lightning detection systems, of weather radars, and satellite data. Interfaces for additional systems, such as radio sonde systems, wind profilers, radiation station, Ozone measuring instruments such as Brewers or Dobsons, etc, can be added easily. Combining data from different sources to produce new products. For example precipitation from a weather station is combined with precipitation from the weather radar. The latter provides information on precipitation in the vicinity of a station, which is used for the improved
WaWa code. The same can be done with cloud information from ceilometers and satellite, and lightning detection data. Communication with Airport Servers. In the MetConsole concept airport servers are identical to the NMI servers, but the only collect data for one airport and usually there are installed at the airport as well (not required). Communication with the Regional Centres Servers. Regional Centres can have their own MetConsole servers to collect data from AWS stations for one region in a country. This data can be used locally, and automatic exchange of data is set- up to forward all relevant data to the NMI Headquarters servers. Data Acquisition from other networks, for example hydrological networks. The data coming from these other networks can also be processed 3.3 Airports MetConsole AWOS systems contain dual redundant servers, clients for the users, and usually 1 or more AWS with the sensors connected to them. The AWOS server configuration is different from an AWS network configuration. For example, the ICAO and WMO have special requirements for typical aviation meteorological measurements for visibility (RVR), pressure, etc. The MetConsole servers have special communication modules through which they automatically exchange data with each other at a nation wide level. 3.3 Clients In Figure 1 many clients are shown. A client is a powerful machine. Once it has a connection to the network, for example via the office LAN or via dial in from home, all systems in the whole network can be accessed (password protected!!!) Clients are used by observers to make METARs, they contain various screens with graphical presentations of meteorological data (configurable), but also software upgrades for all the systems in a country can be achieved with a few mouse clicks. Station configurations are done with clients as well. Basically all user interactions with the systems are done with the clients. 4. Examples 4.1 KNMI, The Meteorological Network in the Netherlands In 2000 KNMI and the supplier of the new network (Almos Systems, now Telvent Netherlands) started the project to implement the new meteorological network. In total there are 23 fully automated observation stations, 5 civil airports and 2 navy airports, 9 air force air abses and a central site, De Bilt, where all data is collected, processed and stored. All meteorological reports are generated at site, and they are disseminated via the message switch (MSS). The ministry which KNMI is part of, also operates a network of some 70 weather stations along the Dutch coast. Data from these stations are collected by a dedicated server (non-knmi), and then they are forwarded to the Dutch meteorological server for further processing, report generation and dissemination, and forwarding to a climatological database. Data from the 2 precipitation radars is forwarded to the meteorological servers, the same for data from the lightning detection system, where both types of data are further processed and finally used in meteorological reports. At Schiphol a wind profiler is also integrated for aviation purposes. In 2005 the network was extended with 9 weather stations installed on oil production platforms in the North Sea. These stations are non-almos stations, but the data is important and can be ingested and processed independent of the make of the weather station.
In the Netherlands all airports had completely equipped meteorological services for 24 for hours per day, both forecasters and observers. At Schiphol, there was a full time meteorological staff of 80 persons. After the airports were equipped with the new networked systems, the airports did not need to have the forecasters on site anymore, as the data was available everywhere. All the airport forecasters were withdrawn from the 5 civil airports to the head quarters in De Bilt, were now only one forecaster is doing the work for all 5 airports. At Schiphol there is now only 1 full time observer left (6 to 8 persons instead of 80). In the Netherlands the 9 Air Force airports are also integrated in the Dutch Meteorological Network. They operate their own servers, which also act as a backup facility for the Met Office servers. If a disaster would happen at the Dutch Met Office and they would be completely incapable of providing meteorological information, the military systems will completely take over without loss of functionality. Vice versa, if the military servers would completely fail for whatever reason, the KNMI systems will keep the military meteorological activities fully operational. The servers include some of the most sophisticated state-of-the-art processing modules that are implemented in the World. For example, the generation of SYNOPs for each WMO station in the Netherlands is done using not only instrumental observation from the station itself, but it is combined with data from different sources: with data from the precipitation radars, data from the lightning detection system, and data from satellite images (cloud information). Similar procedures have been implemented recently for AUTO-METARs at airports. METARs can still be produced by observers but the systems are capable of fully automatically generating METARs, including human observations, and with sophisticated combination of all available additional information. The AUTO-TREND is the next step, so that he airport now-cast will also be fully automatically inserted in the METAR. 4.1.1 Central site De Bilt The De Bilt central site is the heart of the synoptic meteorology in the Netherlands. The data from all station (automatic and airports) is collected here, data is processed here, and several reports are generated here and disseminated via the message switch (MSS) in De Bilt. The role of the components: CIBIL (and its hot-standby) is the new central systems (MetConsole Servers) for the KNMI observations network. It collects observations from all automated stations and airports (every 10 minutes via ISDN, LAN or WAN, being upgraded to GPRS), collects data from the RMI network, which is a non-knmi network with 66 meteorological and hydrological sensors in the Dutch coastal region, collects data from the KNMI lightning detection system (SAFIR) and derives information for the neighborhoods of all stations, collects Meteosat cloud pictures and derives information for the neighborhoods of all stations, collects data from the KNMI precipitation radar (Gematronic) and derives information for the neighborhoods of all stations, it performs all other calculations to obtain the so-called derived quantities, tries to recover missing data (if applicable), fully automatically generates SYNOPS and other reports at the time they are due, fully automatically sends all reports and bulletins to the message switch (MSS), performs quality checks of all data coming in and for the derived quantities,
is used as a powerful maintenance tool to monitor the status of all components in the KNMI observations network (from individual sensors to all servers present at the various airports, etc). CIBIL is based on Windows 2000 Server and the MetConsole software from Almos Systems. The hardware consists of standard commercially available servers. A GPS is connected to CIBIL, which synchronizes all servers, all MIOUs and GDIS locally in De Bilt, and also those connected to CIBIL via the WAN. Figure 2 The typical CIBIL home screen Automation of visual observations With the introduction of the new KNMI observations network, KNMI only still had human observers at the airports. All other stations will be unmanned from the end of 2002. At the moment preparations are being made to remove observers from all regional airports as well. At some 10 main stations throughout The Netherlands, stations have been equipped with ceilometers and present weather sensors. In CIBIL the cloud measurements for each station are combined with Meteosat measurements from clouds. An algorithm has been implemented that derives cloud cover and cloud layers for each station based on the combination of these measurements. CIBIL also acquires data from the KNMI SAFIR lightning detection system and our precipitation radar. In CIBIL this data is combined with the present weather information for each station to generate the WaWa code for the SYNOP for those stations. From 2003 KNMI produced only fully automatically generated SYNOPs that contains information obtained with instruments only. It is expected that in the near future fully automated METARs (and AUTO-TRENDS) will be implemented for all regional airports as well. In De Bilt several additional MetConsole Servers are present: the KMDS is a mirror machine of the CIBIL. Users that want to have data from the real time network connect to the KMDS to retrieve data from this machine instead of the operational CIBIL. The ADS is a database server that stores all the observational data and reports from all airports in The Netherlands for a period of 100 days (based on an ICAO requirement). The GDIS systems, the MetConsole Clients, in De Bilt are identical to the ones at the airports. They can be used to log on to any of the servers, i.e. to CIBIL or to any of the airports servers. Depending on which system one logs on to, a home screen is shown. For the CIBIL this is shown in Figure 2,
The important thing is that one is able to access any of the servers via the LAN or WAN from any site using any of the GDIS. In Figure 2 the typical CIBIL home screen is shown. On the left the various components of the observations network can be seen, the middle pane shown an overview of all servers in De Bilt and the systems presents at the airports and automated stations. On the right all stations are visible on the map of The Netherlands, red indicating that there is something wrong at those stations. For maintenance purposes and trouble shooting it is possible by clicking on the stations (and also on the buttons or network icons) to descend down to sensor level to look at the performances of all systems in the chain. On the bottom right, the alarm list is shown. 4.1.2 Dutch Airports The systems at all airports (7 in total, including Schiphol as the largest) and the central site in De Bilt can communicate to each other by WAN connections. The components at the airports: ADCM The AWOS MetConsole Server. This is a server that acquires and stores the data from all sensors (approximately 60 at Schiphol). This server also computes all derived quantities that are required for meteorological reports, makes those reports automatically at the times they are due, and sends them to the message switch in De Bilt. The ADCM has a hot-standby backup, that takes over as soon as the primary server fails. GDIS The MetConsole Client. The GDIS is a graphical display system that is connected as a client to an ADCM. This can be the ADCM that is located at the airport itself, or one of the ADCMs that is at one of the other airports. The GDIS has 14 different screens to display meteorological information from the airport to observers and forecasters. There are overview graphs, maps of the airport with the RVR, wind, etc. Also the screens for observers to prepare the METAR and SPECI are found here, although the reports themselves are made on the ADCM. The GDIS thus is a multi functional graphic display system. An example of the Schiphol overview map is shown in Figure 3. Clicking on one of the buttons on the right side of this screen, navigates to a new screen. Figure 3 The overview map of Schiphol on the GDIS. MIS The MIS is a server (also with a hot-standby) with only one task: to enable local users at an airport to get meteorological information that is to be supplied by KNMI. 4.2 MeteoSwiss Meteorological Network In 2002 a project started at MeteoSwiss to install a network of approximately 70 automated weather stations in Switzerland. The data from these stations are collected every 10 minutes by servers in
the MeteoSwiss Headquarters in Zürich. When the installation is finished, Switzerland will have a semi-real time meteorological network with all observations available at a 10 minute basis. In Figure 4 below, a schematic representation is shown of the meteorological network, the Swiss MetNet, or SMN, that is currently being installed in Switzerland. Figure 4 The architecture of the meteorological network as it is currently being installed by MeteoSwiss in Switzerland. The main features of the network include: ADAS Stations The new generation weather station based on network technology, supporting Ethernet, LAN and WAN connections with other computer systems. CDAS/NIMDAS The dual redundant servers, responsible for scheduled data acquisition, processing of data, quality checks and forwarding of data to the message switch for report generation. Clients The systems users have on their desks. They connect to the servers to receive data, to change station configurations, to monitor the operational status of all systems in the network. Some clients are located in Zürich, some in Payerne and some are installed on laptops. Diagnostics The server and client software includes extensive facilities for network monitoring and diagnostics. Examples include remote signal monitoring, network overview screens and alarm reporting facilities. Networking Communication systems are at the core of the MeteoSwiss installation. One of the most important features of the system is that they apply full network communication facilities between servers, clients and weather stations. Expandability The software enables a wide variety of server systems to be interconnected together, including major airport server systems and individual manned observation stations, as well as central systems such as the CDAS/NIMDAS systems. 4.2.1 Possible Expansion of the MeteoSwiss Network Looking at the current situation on the meteorological systems in Switzerland and the MeteoSwiss plans for the future, it is possible to integrate the observation system to a large extent. The basic modules that can be added to the current configuration are Airports, separate network, but can also be fully integrated in the SMN. Precipitation radar, data can be integrated to provide value added products, Radio sounding (normal and ozone), Nuclear power plant meteorological stations, CN-MET, separate as a sub-net, Future instruments such as wind profilers, LIDAR, microwave instruments, Other networks
4.2.2 Precipitation Radars, Lightning Detection System, Radio Soundings, Ozone measurements, water vapour LIDAR There are 3 precipitation radars in Switzerland, Monte Lema, La Dôle and in Albis, see Figure 5. They all have there own processing computers and processed data is available from those computers. These computers can easily be integrated into the automated Swiss MetNet. Processing of this data can be similar to KNMI, i.e. for improving the weather code (WW or WaWa) for the SYNOP. If a lightning detection system is present the same procedure applies to that type of data as well. In addition it is possible to display the radar data/lightning detection data (or composite images) on all the clients as well. Because of the network structure of the servers, all the data, derived data, computed and processed data from one server, will be real time available on all other servers and clients as well. For example, precipitation and lightning information computed in Zürich would be real time available at any airport to be included in METARs and SPECIs. Although not discussed any further here, radio sonde data can be acquired by the servers, processed for display purposes and further dissemination. In Arosa, Dobsons and Brewers are installed for monitoring total ozone. The output files produced by the PC controlling these instruments can automatically be acquired by the central servers and further processed, or disseminated if necessary. In Payerne a water vapour LIDAR will be installed. Controlling the LIDAR PC can be done remotely (for example from Payerne) using standard tools provided in computer operating systems or with special software designed for remote control of computers. The data can be ingested and processed for further use by the CDAS/NIMDAS. 4.2.3 Nuclear Power Plant Meteorological Stations, CN-MET There are 4 nuclear power plants in Switzerland, their locations are shown in Figure 5. MeteoSwiss is planning to set up 3 meteorological stations in Payerne, Wynau and Schaffhausen, with additional turbulence sensors (3D wind), wind profilers and μwave (water vapour), to support analysis of dispersion of nuclear material in the atmosphere in case of an accident at one of the power plants. These sensors are providing data to the SMN either via a regular weather station or via their dedicated control and acquisition PCs. The data coming from these special stations can directly be forwarded to the systems that are used for running the dispersion model for input into the model, comparison of the model with observations or further analysis. 5. Summary and conclusions The concept of a completely nation wide meteorological network contains: Automated stations, Airports, both civil and military, Standard backup facilities, Other networks providing meteorological information (coastal station, agriculture, environmental, climatological stations, hydrological stations, etc), Special stations (nuclear power plant, electrical power plants, other industrial stations), Network of cameras, Precipitation radar networks, Lightning Detection networks, Research weather stations,
Radio sonde stations, Other meteorologically related stations or special sites with LIDAR, wind profilers, etc. Turbulence Wind Profiler μwave Turbulence Wind Profiler μwave Albis Precipitation Radar Radio Sonde LIDAR SODAR Turbulence Wind Profiler μwave Arosa - 3 x Brewer - 3 x Dobson La Dôle Precipitation Radar Monte Lema Precipitation Radar Figure 5 Precipitation radars, radio sondes and stations with rare measurements, such as μwave, LIDAR, wind profiles, turbulence, Brewer and Dobson, and SODAR. The nuclear power plants in Switzerland are also shown. In Arosa 3 Dobsons and 3 Brewers are present for measuring total ozone. As pointed out in this paper, many of the meteorological observation systems can be integrated in real nation wide network with all the benefits that emerge from such a system: Possibility of centralization of maintenance, Possibility of centralized meteorological operation, Cost reduction, High reliability, A standard interface for all systems for all users and maintenance staff. It is expected that integration of the observations systems will be the main development in the next generation MOIS, where also models and GIS applications will be included for improved real time validation of the observations. References The new meteorological observation network in the Netherlands, F. Kuik and T. Haig, paper and presentation at the AMS 2002, Orlando, 2002. SwissMetNet: Renewal of the Swiss Meteorological Networks, Heimo, T. Konzelmann, B. Calpini, J. Rast, N. Tschichold, E. Grüter, paper and presentation at the TECO 2005, Bucharest, 2005. The new meteorological observation network in the Netherlands, Status and operational experience, W. Wauben and D. Hart, paper and presentation at the TECO 2005, Bucharest, 2005. Meteorology and Security around the nuclear power plants in Switzerland, B. Calpini, J-M. Bettems, and D. Ruffieux, paper and presentation at the TECO 2005, Bucharest, 2005.