The cost for developing and maintaining the databases can be found in Table 7. It can be seen that it is cheaper to use and maintain a commercial database, however, factors like speed of access and ability to store complex data types have not been addressed in this study and should be considered before deciding on the best option for a particular system.

The reasons why countries have developed their own databases have not been looked at in this study.

There are 31 countries that have developed their Forecaster Workstation System themselves. Of these countries, 20 of them have developed more than 70% of the system themselves, 3 have developed more than 70% of the system from free 3^rd party software, the rest were a mixture of free 3^rd part, commercial software and in-house applications. The most popular languages for developing the Forecaster Workstation System are C (28 users) and Fortran (26 Users), although C++ (16 Users) and Java (10 Users) were also used by a number of countries. The other languages that participants used were Visual Basic, Perl and Delphi but these were not used extensively. Few countries used any other application from NMS’s or WMO, two countries use MAGICs from ECMWF and three countries use the PCGRIDDS package.

There was no obvious difference in purpose between Unix and PC based systems. There are 60 systems included in the analysis of the purpose of the Forecaster Workstation System. Of these, 21 systems are used only for visualisation of meteorological data, 3 are used for production only and 36 are used for both visualisation of meteorological data and production of charts for both internal and external customers. The details for each WMO region are given in Table 3.

Most of the systems visualise a variety of data, NWP model data, Radar and Satellite images and Observations. The most popular NWP models used are UKMO (24 Users), ECMWF (23 Users) and NCEP (15 Users). Of the 60 systems, 17 use their own global model and 30 use their own local area model. Most of the systems make use of observations from the GTS (51 Users) and supplement this with local observations (47 Users). Just over half the systems have some form of Imagery (radar and Satellite data). Local imagery is used by 33 of the systems and images from other NMHSs is used by 31 systems.

The majority of the systems use WMO formats (50 systems) the most widely used are GRIB (38 Users), BUFR (18 Users), GRID (14 Users) and OPMET (14 Users). This allows for a good interchange of data between countries, all countries should be encouraged to use standard formats whenever possible.

The Forecaster Workstation Systems are used for a number of different types of forecasting. The most popular use being for general forecasting (53 Users), other types of forecasting done on the systems are Aviation forecasting (36 Users), Severe Weather Forecasting (31 Users), Marine forecasting (26 Users) and Long Range Forecasting (12 Users). The details for each WMO region are given in Table 4. We could surmise from this that the development of tools to visualise data for the safety and protection of people and property is one of the main drivers behind the development of the Forecaster Workstation System.

The availability of more sophisticated tools was analysed, 35 systems, just over half, have the ability to integrate and manipulate satellite imagery, NWP data and meteorological observations. Some systems then allow for the development of forecast products to take place on top of these overlays, for example the production of Significant Weather charts for Aviation. There are 43 systems which have a forecast preparation tool as part of their functionality, This allows for the addition of Manual text (31 Users), Manual Graphics (26 users), First Guess Text (20 Users) and First Guess Graphics (17 Users). The use of Manual Text and Graphics is more advanced than the introduction and use of first guess data. The exploitation of first guess data is potentially an area where countries could collaborate on the future development of these techniques.

There are 30 questionnaires used for the analysis of future developments, Table 8 gives details broken down into WMO Region. The three operating systems that will be used in these systems are Unix (11 systems), Windows (11 systems) and Linux (8 systems). Only 28 of the systems gave more details on how they hope to develop this new system. There were 13 new systems, which would be bought in as complete systems and 8, which were going to use a mixture of bought in systems and in-house development.

There are 7 systems, which will be developed completely in-house. The reasons for this were, not being able to afford to buy a system in, more flexibility to adapt the system to the needs of the individual Meteorological Service and the requirement to build on existing systems and to keep expert knowledge with the organisation.

There were 10 countries that do not have a Forecaster Workstation System at present. There were 3 who planned to acquire a system in the near future, two were going to purchase a complete system from another NMS and one had not decided. It was considered important to have a Forecaster Workstation System to enhance accuracy of forecasting and expansion of professional skills. The cost of such systems was seen as the main reason why one had not been procured, although the lack of trained IT staff and other specialist in the specific organisation or country was also seen as a potential problem for developing or acquiring such systems.