7. RECOMMENDATIONS AND PROPOSED WAY FORWARD

The ad hoc groups reported their preliminary findings, and with some modifications in their topics, were invited to reconvene to prepare final recommendations.

7.1. Ad hoc Group 1

The first group developed a few specific proposals for revising the JDIMP terms of reference (see Annex III), and provided a framework to discuss some of the procedural issues facing the panel.

The group reviewed the GOOS DIS concept whereby day-to-day data issues would be addressed. Such a group is needed since the modules of GOOS have very disparate data content which requires special consideration. With regard to GTOS, it was noted that the information will mainly reside with networks (of sites) which will be under the control of a Network Panel. A DIS-type concept is not envisioned at this time. Regarding GCOS, it has relied upon the work of DIMP to address its data issues. This special function toward climate data, since it is the integration of elements of atmosphere, ocean, and land, should remain a special purview of JDIMP. On the basis of these observations, the group was able to formulate a ‘strategic’ concept of JDIMP whereby it would identify common elements among the G3OS and provide an overarching data management structure support to them. It was particularly noted that JDIMP should look to the integration of space-based and in situ observations for the three.

7.2. Ad hoc Group 2

In reviewing the overall G3OS plan, the group first agreed that the concept of a “data system” is still valid in that it provides a vehicle for cooperation and linkages, and can contain regional and global elements. However, it advised that the plan should not be ‘rigid’, but accommodate changing needs. It determined that a substantive revision is needed and a simple substitution of ‘G3OS’ for ‘GCOS’ is not acceptable. It recommended that the plan clearly differentiate between ‘data’ and ‘information’. The group reviewed the current plan and made comments chapter-by-chapter:

• Chapter 1

  Stress linkages to the science programmes;

  Provide for feedback from users.

• Chapter 2

  GCOS data and information principles (agreement with GOOS & GTOS ones);

  Add a new scenario that cuts across the 3 programmes;

  Give an example of feedback in one of the scenarios.

• Chapter 3

  Add section on system performance controls including:

  - monitoring and feedback,

  - linkage between system components,

  - note and correct deficiencies in system (e.g., Does calibration support the data? Has a station stopped reporting? Do data go on the GTS and vanish?),

  Emphasize linkages:

  - extend the GHOST concept to the data system,

  - link small, regional centres and larger global centres in a hierarchy of centres.

· Chapter 4

Add periodic system reviews.

· Chapter 5

Consider the need for this chapter.

· Annex I

Prepare a new three-year plan;

Develop separate plans for each component;

Emphasize linkages.

· New Annex to describe where we are now, to include information on existing elements.

The group initially recommended that the three project offices prepare a revised draft Plan for submission to the next JDIMP meeting. However, after some discussion, the JDIMP Chairman and Mr Keeley agreed to contribute with the project office assistance and support.

The group also considered the role of the “Information Center” (IC) and determined that it should be a pilot activity whose goal is to evaluate the IC as an effective source of information for the G3OS, and as a potential long-term G3OS element. If evaluation is positive, the IC should be transferred to an institution that can provide long-term support. During the pilot phase, the IC will:

• Focus on information systems;

• Point to other centres and information systems;

• Provide access, but will not hold data,

• Present information as held in the centres, without either creating or modifying the presentation.

The group agreed that to meet the diverse needs of the G3OS needs, the IC must have considerable scientific input and collaboration with the GOOS and GTOS communities.

The participants discussed the development of the plan as recommended by the group, and agreed with its assessments. The project offices will be asked to develop a timetable to prepare Version 2.0 in a timely fashion so it may be reviewed by members of the panel. The participants also agreed with the need for an IC. Further, it was agreed that Dr Webster would revise the existing document about the IC, and submit it to the JPO for distribution to the principals of the G3OS for comments and advice. The JSTC-VII meeting in September was recognized as the optimal opportunity for discussion and final resolution by the principals. Dr Spence agreed to have the proposal tabled at JSTC.

7.3. Ad hoc Group 3

The group decided first that it was important to define what is meant by the term “metadata” and proposed the following definition which it felt was broad enough to be applied to all G3OS data sets:

Metadata: The information about the data contained in data sets which is needed to understand the content and optimize the usefulness of the data set.

For example, metadata should answer questions such as:

- What are the elements observed?

- What is the station name and type?

- What is the current location? Has there been a relocation?

- What is the site description?

- What are/were the periods of observation?

- What instrumentation is/was in use?

- What are/were the observing practices?

- What conversion algorithms are/were used?

- Is it original (raw) data? Is it quality controlled? Is it homogenized?

- What code/format was used? What code/format is being used now?

- What is the availability/access of the data?

The group felt that it was not appropriate for JDIMP to get involved in trying to determine the quality of data sets or to ‘certify’ data sets but proposed instead to ‘register’ data sets based on the available metadata associated with each data set. The data sets would be selected from the domains of the G3OS but should not be tagged as being associated with any one of the G3OS. Also, data sets that are not global in nature but are of global interest may be included. In developing this registration procedure, existing cataloguing and data set referencing systems such as INFOCLIMA should be taken into account. This proposed data set registration process was viewed in three stages or levels:

1. Creation of a high-level catalogue system for G3OS data sets on the World Wide Web;

2. Development and inclusion of a guide to potential users of the G3OS data sets;

3. Development of a set of guidelines for the preparation of detailed metadata for G3OS data sets that would be held with the data sets.

The group reviewed the Principles of Environmental Monitoring that were presented to the meeting on In Situ Observations for the Global Observing Systems held in Geneva from 10-13 September 1996. They agreed that they were important and could be promoted through JDIMP. Several minor modifications were proposed for a few of them (see Annex IV).

The group made the following recommendations:

1. JDIMP should adopt a strategy to register selected data sets. As part of this process, metadata guidelines for G3OS data sets should be promulgated to specify metadata at the data directory level, provide basic information for a “guide” to the use of the data set, and promote metadata rescue and the development of comprehensive metadata on G3OS data sets which would be referenced in the “guide”.

2. JDIMP should further review the revised Principles of Environmental Monitoring in the context of the G3OS and then combine them with existing data management principles to include such principles as promoting the long-term archiving of climate data in digital format.

The group developed a pilot project to test the feasibility of registering selected existing data sets based on the development of a directory level of metadata and a “guide” to the use of the data set. The guide would be based on the results of the most recent questionnaire contained in the conceptual model on documenting climatological data sets.

Mr Searle offered to be the focal point for the pilot project which will be based on an existing system in use in Australia. He will review a number of existing directory level metadata formats and propose a recommended format to be used for the G3OS. He will make necessary modifications to the existing software that would be used to catalogue existing G3OS data sets at this directory level format. Mr Scholefield will distribute a package of information describing INFOCLIMA, particularly its directory entry level of cataloguing to the participants. Potential G3OS data sets include those on a listing produced by the last DIMP meeting in Ottawa. Messrs Heino, Zhai, and Razuvaev each offered to provide input on a data set with which they were familiar. Another possible candidate would be the GSN now being established. Mr Searle will provide each member of the group with a copy of the software (based on Microsoft ACCESS software) so they can use it to input appropriate data set directory information and return the completed information to him on a diskette or by e-mail. He will put the results of the pilot project on the existing “Blue Pages” information system in Australia [URL: http://www.aodc.gov.au]. Detailed actions pertaining to the first component of the data set registration process and a schedule are listed below in Table 1. Regarding the second component, Mr Scholefield will get the most recent version of questions from the conceptual model for the guide and distribute them to other participants in the pilot project to see how it might be incorporated into the pilot project.

The plenary discussion was very supportive of these specific activities. Participants pointed out that there are two levels of metadata, those that: 1) characterize the datasets for users to obtain an overall concept of their contents (i.e., directory level); and (2) characterize the data at the level of the observations themselves (i.e., data content level). The latter information would not be appropriate in the IC or other directories, but would be essential when examining the data for specific uses. The project above and the IC address level 1).

Table 1
JDIMP Metadata Pilot Project Implementation Schedule

In addition to the explicit proposals developed in the ad hoc groups, Messrs Hasegawa and Keeley developed a recommendation related to the NEAR-GOOS and IGOSS/IODE activities. In outline, it was observed that since:

i) IGOSS/IODE, through the GTSPP, collects and distributes temperature and salinity profile data (see 4.7) and

ii) NEAR-GOOS expects to incorporate more real-time and delayed mode profile data of interest to GTSPP (see 3.2);

JDIMP should request that GTSPP and NEAR-GOOS (recognizing project differences) explore points of common interest to increase cooperation and decrease potential duplication of effort.

The recommendation was made that funding be solicited for a workshop to be held in the South-East Asia region to examine end-user data, information, and products for environmental scientists and policy-makers, particularly in key socio-economic sectors. No specific action was identified to advance the proposal.

7.4. Other Issues

In plenary discussions, several overarching issues were discussed including a ‘template’ to guide the work of the JDIMP. The template (Figure 1) outlined the principal stages in the development of products. The sequence of stages included:

1. Development of the science issues based on political drivers and conventions;

2. Identification and specification of variables to be observed;

3. The methods of collection, calibration, and quality control and data consistency;

4. Mechanisms, formats, and codes for transfer of the data;

5. Processing of observations, including gridding and analysis;

6. Archiving of data, information, and products as appropriate; and

7. Development of end-user products.

Throughout the template, responsibilities for oversight are noted. The final element also includes the role of evaluation of the overall process including problem solving, feedback, and documentation.

It was agreed that (1) and (2) should not be the purview of the JDIMP, but that elements of (3), (4), and (6) definitely should be. Knowledge of items (5) and (7), while not the primary concern of the panel, should be considered to provide a comprehensive approach to information management. In addition to considering this paradigm for JDIMP, the panel agreed to consider its specific oversight roles - i.e., How does the data and information process really function? What can and should be done to ensure effective management of the information? What constitutes adequate documentation?

The plenary session also considered the membership of the panel and related administrative issues. The JPO Director provided a description of the process to develop the list of participants for the meeting: (1) the original membership of DIMP was reviewed and those able to attend were invited; (2) the GOOS and GTOS were invited to recommend individuals; (3) the chairmen of joint panels (AOPC, OOPC, TOPC, GOSSP) were invited to attend or designate representatives, and (4) representatives of other relevant organizations were invited. Although the procedure was sound, the intended balance was not completely achieved. Representation from GTOS and GOOS should be strengthened, and the number of former DIMP members should be diminished. The representation from related organizations should be ‘cycled’ to enable liaison with them. It was agreed that the final list of members would be designated at the JSTC meeting where joint G3OS panel issues will be addressed.

It was agreed that the next meeting should be held in spring 1998.