Policy Committee Meeting Summary
18-19 May 1998

Narragansett, Rhode Island

Participants

*Members*	*Representatives*
John Merrill (Chair)	Otis Brown (UCAR Trustee Liaison)
Robert Fox	Harry Edmon (ATAC)
Colleen Leary	David Fulker (UPC)
James Moore	Bernard Grant (NSF/ATM)
Charles Murphy	Clifford Jacobs (NSF/ATM)
Julie Winkler	Mohan Ramamurthy (U. Ill./Users)
	Tim Spangler (COMET)
*UPC Staff*
Sally Bates
Ben Domenico
Joanne Graham
Jo Hansen
Linda Miller

Administrative Matters

The summary of the last meeting should be amended as follows: Under the NOAAport discussion, the last discussion point should read:

The next meetings of the Policy Committee will be:

Status Reports

Director's Report and the Budget Report

Fulker gave the Policy Committee the same presentation he made to the NSF panel reviewing the Unidata 2003 proposal. Copies of his presentation slides and copies of his normal status report are in the notebook. Fulker also gave a verbal report on his trip to Paris for a WMO meeting. He reported that the primary topic of discussion was the NWS's decision, announced at the meeting, to place all GTS data available to it on the Web, regardless of some countries' wishes to restrict access to their data. The origins of this decision are not clear, and it was greeted by the Europeans with dismay. Copies of the slides from Domenico's presentation are in the notebook; he noted that there would be no carryover at the end of this fiscal year.

Discussion

The uncertain future of OS/2 makes Unidata's move to Java even more important.
NIDS: the cost of NIDS was not a topic, probably because the panel accepted UPC's explanation that Unidata has no control over this issue. The Policy Committee expressed concern about the lack of tools for manipulating NIDS.
Access to retrospective data is an important issue to users as well as the review panel; however, its scope may be enormous, entailing issues of policy, resources, and servers. UPC is already 'nibbling at the edges' of this issue (e.g., the creation of the SSEC image archive, and SCD's creation of

Fulker was questioned whether the WMO viewed the LDM as a solution to its Y2K problem
The Policy Committee expressed concern that the adoption of the LDM by other countries might create support problems for Unidata.
COMET has three more years of funding for AWIPS training and plans to create more case studies.
Unidata's financial status beyond FY 1998 is uncertain.

Users Committee Report

Ramamurthy reported on the March Users Committee meeting, noting that the committee had failed to address the question of a topic for its next summer workshop. A copy of the meeting summary is in the notebook.

Discussion

The chance to meet with NWS's Tim Crum was particularly useful; NOAA personnel at his level rarely get direct input from the university community.

MetApps Report

Murphy described the history of the nascent MetApps Task Force, noting that the group was still short two members and was planning to meet 20-21 July in Boulder to design the groups communications structure.

Discussion

UPC may want to consider adding liaisons to the group from COMET and ATD and perhaps even a federal lab; these groups are also involved in creating new meteorological applications.
UCAR's IT initiatives might impact the task force's activities.
There was concern that the task force looked like a design specifications group, which suggested that UPC was engaging in a full-blown software development effort.

NOAA Report

Fox distributed a copy of the NOAA report based on his discussions with Doug Sargeant.

Discussion

NOAA is funded through AWIPS build 4.2; future enhancements are being pursued.

NSF Report

Grant reported that the UCAR Cooperative agreement has been approved by the National Science Board after extensive review. UCAR's funding for FY 98, however, was reduced by 1.6% ($0.95M) even though, when compared with other NSF-funded organizations, UCAR is one of the best (e.g., the astronomy centers are being re-competed). Grant also reported on Geo's goal of developing a long-range (10-year) vision. The vision document is scheduled to be published in the Fall of 1999.

Jacobs reported on KDI (he is Geo's representative to the KDI Working Group). The group has established some procedures: proposals spanning two or more KDI programs (i.e., KN, LIS, NCC) will be reviewed by two or more panels; more reviewers can be added (i.e., mail reviews) if needed; awards must be made by the end of the fiscal year. There were 1150 letters of intent, but KN received 256 proposals, NCC 236, and LIS 49. NSF anticipates funding 60-75 of these. Panelists will convene in two sessions: July 27-29 and Aug. 3-5; there will be 10 parallel panels in each session. Jacobs will be submitting the Unidata: 2003 proposal for consideration in the competition for "extant" proposals.

NSF Review Panel Recommendations

Fulker presented and the committee discussed a draft response to the recommendations made by the NSF review panel. A copy of his initial draft was distributed at the meetings. In the course of the meeting, additional ideas for responding to the panel's recommendations were expressed. These were:

Recommendation 1: Unidata is pleased that the panel recommends continuing the program.
Recommendation 2: [see Unidata Budget section below]
Recommendation 3: Mention DODS initiative; need to incorporate UCAR IT strategy and access to NCAR/SCD data sets
Recommendation 4: Mention that PAGE wasn't in existence at time of proposal
Recommendation 5: The potential use of LDM on inexpensive platforms by WMO members world-wide might pose a support problem.
Recommendation 6:

Attachment: UPC intends to talk to Oklahoma, which is establishing a 9-radar network to gather Level II data; could consider this a prototype Level II IDD system; will need tools; will investigate collaborating with ATD on analysis tools for radar data.
Discussion: Cost to ATD of "finishing" Zebra; need more information about the magnitude of the effort before committing to this work; need to balance need for retrospective data against the move to Java; the committee asks that Unidata be clear of the relative priorities of radar vs. other forms of data. ***The committee noted that the statement in Finding 6 that there is a convergence of needs and technologies is NOT true.***

Recommendation 7: Equipment grants will be particularly important in the effort to help sites transition away from OS/2.
Recommendation 8: NOAA has issues about network security (firewalls) and policy questions; the ups will try to find ways to save money, but the program still needs to budget for downlinks.
Recommendation 9: It is not clear what is really meant by a "spreadsheet environment." The ghost of Ynot lurks.
Recommendation 10: This is already being done by the staff; the UPC will simply publicize the pages.
Recommendation 11:

Resolution 1:

The Policy Committee tasks the UPC to investigate methods to improve access to historical data as specified in Recommendation 3 and Level II radar data as specified in Recommendation 6 and report back to the Policy Committee at its next meeting.

Demonstration of the Distributed Ocean Data System (DODS)

Peter Cornillon ( U .Rhode Island) gave a demonstration of DODS. He noted that it's a UNIX-based system designed by oceanographers to enable them to serve data to each other; however, there is nothing inherent in the software to prevent it from being used to serve others kinds of data. Essentially, the software facilitates how a users subsets, acquires, and ingests a data set via a URL. DODS is built on the client/server architecture and requires that the data reside on a DODS-compliant server. To date, DODS servers can handle data several formats (netCDF, JGOFS, HDF, MATLAB, FREEFORM, DSP). The DODS developers are not yet pushing the community to populate DODS servers since the GUI needs more work.

Discussion

Is similar in many respects to SSEC's ADDE server; perhaps SSEC and Rhode Island should confer on connecting the two; might use McIDAS to look at ADDE/DODS data.
Populating DODS servers might be a problem--a lot of metadata need to be created. While the data can be (and is being) served without the metadata, the GUI, which facilitates the use of data, requires some metadata, and obtaining these requires work. The UPC would probably have to take on the effort of populating servers initially; users would help only after being convinced that the system works.
The GUI that the DODS group has developed greatly facilitates the use of data, but it works with a select set of data at present. To make the GUI capable of handling a much broader array of data, such as currently handled by the LDM, will require more work.

Action 1:

At the next Boulder meeting of the Policy Committee, the UPC will invite a representative from SCD to report on university access to data stored by that division.

Unidata Budget

In the context of discussing the NSF review panel's recommendation 2, it became clear that there was some confusion about the tack the UPC should take in modifying its proposed budget. Jacobs explained that any program receiving $3M per year or $15M over five years is required to be internally reviewed and then reviewed by the National Science Board. Such reviews are very unlikely to result in increased funding and might invite comment on UPC activity with unpredictable consequences. Therefore, Jacobs will not allow base funding for Unidata to approach this amount. Unidata should plan for funding to remain at its current $2.3M for the duration of the proposal period, adjusted only for inflation. However, revised budgets submitted in response to the review panel comments can indicate slightly higher than inflation increases. Jacobs DOES have the flexibility to make supplemental awards as needed for specified activities; this does not require a full-blown proposal, but an addendum with a page or two of explanation and a budget request for the activity. In recent years, Jacobs has funded a number of activities for Unidata by this method.

Discussion

There was considerable concern expressed about requiring the UPC to plan based on its current base funding of $2.3M/year. Since the program has been spending at a higher rate in recent years, there was doubt that the program could function at a base of $2.3M.
Jacobs was asked whether the reviewers gave any indications on what activities could be cut, and Jacobs responded only that the recommendations were in priority order.
The UPC could respond to the panel review in either of two ways: (1) to formally close the proposal process, respond as though a large budget might be possible and then handle the cuts as a separate process or (2) respond to the recommendations within the context of the smaller budget.
Unidata is both a development activity and a utility; the UPC needs to know the relative costs of these two activities.
Unidata cannot be just a utility--it must be an innovator as well.
The technologies are evolving so rapidly, it is impossible to set priorities past two years in any event; Unidata may not have a problem in two years--someone else may have created the needed technologies. Therefore, the UPC should identify one or two activities to focus on in the next year or two and then reevaluate its direction and needs.
The future may hold land mines as well as treasures: e.g., the end of the cooperative agreement between NSF and MCI for the vBNS may have significant impact on everyone.
In adjusting to the budget reality, Unidata must ensure that data continues to flow (i.e., the UPC cannot drop FOS contract until NOAAports are in place and running reliably).

Resolution 2:

The Policy Committee thanks the NSF review panel for the efforts, diligence, and insight shown in its Unidata Program review. The Policy Committee gratefully accepts the recommendations of the review committee and will consider each and every one of these in establishing and managing Unidata priorities for the next five years. The Policy Committee will treat the recommendations to be prioritized by the review committee in the order given.

Resolution 3:

The Policy Committee recognizes that the facts of life are a modest increase from the FY 1998 NSF Unidata budget of $2.3M for FY 99, with similar annual growth through 2003. This differs significantly from the $3.643M requested for FY99 in the Unidata 2003 proposal. This modest increase in the FY99 budget over FY98 and NSF's forecast for annual growth will require drastic reductions in the UPC budget. The Policy Committee will work to these numbers until there is a reason to change.

Resolution 4:

The Policy Committee recommends that, in making the necessary cuts to its program, the UPC be cognizant of the vital role that data flows and user support activities play in research and education in the atmospheric sciences.

Action 2: