Policy Committee Meeting Summary:
23-24 January 1997

Miami, FL

Participants

Members	Representatives	UPC Staff
Otis Brown (Chair)	Harry Edmon (U. Wash./ATAC)	Sally Bates
Ken Crawford	David Fulker (UPC)	Ben Domenico
Robert Fox	Bernard Grant (NSF/ATM)	Jo Hansen
Colleen Leary	Arthur (Fritz) Hasler (NASA)	Sandra Nilsson
John Merrill	Clifford Jacobs (NSF/ATM)
Jim Moore	Bill Pennell (UCAR/UOP)
Julie Winkler	Mohan Ramamurthy (U. Ill./Users)
	Tim Spangler (COMET)

Administrative Matters

The summary of the October 1996 meeting were approved.
The next meetings of the Policy Committee will be:
29-30 May 1997, in Boulder, Colorado
25-26 September, in Arlington, VA
A new chairman for the Policy Committee needs to be appointed. Brown is now a UCAR Trustee and will be the Trustee Liaison to the Policy Committee; the May 1997 meeting will be his last as chair.
Harry Edmon is ATAC representative to both the Users Committee and the Policy Committee; one of these roles should be assigned to someone else.
Interesting information: Edmon is sending NIDS Level II data over the IDD; it's on the order of 10 MB/6 minutes, which compresses down to between 2.5 and 0.5 MB/6 min, depending on the weather. The current LDM is capable of handling this.

Action 1:
Nominations for a new Policy Committee chairman should be sent to Brown or Fulker as soon as possible.

Action 2:
The ATAC needs to find a representative to the Users Committee. Harry Edmon will continue as the ATAC representative to the Policy Committee.

Status Reports

Director's Report
Copies of Fulker's Director's Report and Nilsson's budget report are in the notebook. As part of the Director's Status Report, Domenico gave an update on the status of IDD. He noted that:

the system is holding steady;
data delivery is at acceptable levels;
the network problems that do appear are associated with upgrades to the infrastructure;
a new alternative source for data is now in place (UCAR's IITA group), which uses a satellite receiver to supplement the Alden feed; it's missing the RUC data in FOS, however;
the old monitoring system is not being upgraded since staff time is being devoted to decoders instead;
UPC is scheduled to have a latency monitoring system in place by the next Policy Committee meeting, which will include a measure of maximum latency experienced by the community.

Discussion

There was considerable discussion on the request to the Users Committee for Unidata to support the LINUX operating system. The trend toward a multiplicity of possible platforms, applications, and operating systems was of great concern; among the ones mentioned were the WFO Advanced workstation, the request to support LINUX, COMET's GARP, Unidata's move toward Java, SSEC's incorporation of Vis5-D into McIDAS. These pose both technical (how do Unidata applications fit?) and policy issues (what does Unidata want to support?). The community needs and wants a more holistic environment but various components (UPC, NCEP, NWS, COMET) seem to be diverging. For example, if COMET moves to WFO Advanced, how can it continue to support universities (which are not moving to this platform)?
NCEP is opposed to the NWS shift to WFO Advanced; it prefers GARP. This poses issues about configuration and data types. Mention was made of forming a Unidata/COMET task force to discuss this situation with NCEP.
There was considerable discussion on the budget in terms of Java development resources. While Java implementation proceeds quickly, testing an implementation on all the platforms supported by Unidata takes time. Leverage the training costs of learning Java by combining with COMET was mentioned.
The IDD monitoring statistics may be of interest to NSF CISE; the statistics might be helpful in planning for the next generation Internet; some universities are using them as a diagnostic tool for measuring the behavior of the network.
NCAR's SCD and CGD have engaged in an extensive analysis of visualization tools; they may be settling on NCAR Graphics and Windows NT.

Action 3:
Representatives from Unidata and COMET will arrange to meet with NOAA personnel (e.g., Uccellini) to discuss the GARP/WFO Advanced issues.

Users Committee Report
The Users Committee has been focused on planning the summer workshop. the draft program is in the notebook. Ramamurthy noted that the COMET faculty course that was to precede the workshop has been cancelled.

Discussion

There was discussion about the feasibility of inviting secondary-school teachers, among others. There are no restrictions to participation, but the workshop is targeted to undergraduate education. Invitations will be issued to NSF Geoscience personnel such as Bob Ridky.
A second education floater has been added to the McIDAS data stream.
The GEMs program announcement will be reissued shortly after the Policy Committee meeting; the processing for the grants has been streamlined and NSF expects to make the awards this summer.

NOAA Report
Bob Fox spoke with Doug Sargeant and summarized his conversation as follows:

GOES-8 had a mishap a couple of weeks ago. The anomaly appears to have been caused by a failure within the wheel drive electronics for momentum wheel 1. This caused the loss of momentum wheel speed data to the attitude and orbit control electronics (AOCE) computer. The spare was switched into operation and operations appear normal since that time. The loss leaves GOES-8 with no momentum wheel redundancy.

GOES-10 is scheduled for an April launch. It will be checked out and put into "hot" storage or until some other strong need arises. Basically, Wallops Island can only handle two operational geostationary vehicles.

AWIPS is still trying to get congressional and OMB approval to fully deploy the system. They have nine AWIPS Build-1 systems installed in various locations now which have passed pre-deployment evaluation. Deployment has been imminent for some time, but miscellaneous delays on a weekly or shorter basis have taken them into the time frame of a new Secretary of Commerce. The new Secretary will now have to approve and Sargeant has no idea how long this may take. The extent of final deployment is still being debated.

It has been incredibly difficult to achieve satisfactory performance with NOAAport. NOAA is still experiencing unacceptable losses at some sites even after switching to C-band. The system is meeting minimal specifications for long-term bit-loss rate (~10 superscript -6) but performance is unacceptable, especially concerning grids. It appears that the problem may be with antenna side lobes.

The current PRC plan calls for putting lots of radar data on NOAAport. If this comes to pass, there would be no restrictions on anyone availing themselves of these data.

NASA Report
Fritz Hasler reported that an image of hurricane Luis created by his group made Life magazine's annual "big picture" edition. Hasler then demonstrated images of his group's interactive spreadsheet application, which runs on an SGI. They are also able to run SSEC's Vis-5D interactively. Hasler is involved in a joint program with NCAR and Japan to test a distributed remote visualization/analysis application. The project is to test the spreadsheet's use of a high-speed, Internet connection and involves servers at NASA and in Japan, with NCAR running the client. He reported that NASA planned a series of demos at the AMS meeting in Long Beach.

NSF Report
Cliff Jacobs reported on the state of the NSF budget. Copies of his slides were distributed at the meeting. Among other topics, Jacobs said that NSF would convene a panel of 6-8 experts to review the hard-copy version of the Unidata proposal (to avoid conflict of interest, no current member of any Unidata committee may be a reviewer); he is assuming that a separate review by the National Science Board will not be required. Jacobs also reported on a new initiative within NSF called KDI (Knowledge and Distributed Intelligence), which has building "knowledge networks" as one component. This is essentially an effort to build the next generation of networks and databases, converging communications and information technologies. This effort may represent a source of funds for Unidata.

Discussion

KDI is an effort to build on the success of the Internet as a model for technology transfer. How Unidata might be involved was not clear.
Committee members should look at NSF's new merit-review revisions; NSF would appreciate comments. It's available at: http://www.nsf.gov/nsb/merit2.htm

UOP Education Initiative

Bill Pennell reported on a new education initiative being considered by the UCAR Office of Programs. This is in response to the Users Committee resolution last September (1996) and the resolution in October 1996 passed by the AMS Heads and Chairs committee. UOP will be proposing to build on the existing resources of COMET and Unidata. There is considerable university interest in COMET-developed content material and case studies, material COMET already has planned to move onto the Web; the initiative would help speed this process. The initiative would follow the Unidata model--i.e., it would be university driven and adhere to the credo to undertake no tasks that can be done better in the universities. Some ideas had been discussed with Jacobs and Mike Mayhew at NSF; in addition, Pennell held a brainstorming session attended by John Snow (Oklahoma), Don Johnson (Wisconsin-Madison), Gene Takle (Iowa State), Steve Mullen (Arizona) and a number of UCAR staff members. A needs-assessment proposal to NSF is now being drafted by Unidata and COMET.

Discussion

Copyright issues will be a problem; this is already a topic of much discussion on campuses, which UCAR might find helpful.
Unidata's role in the initiative is limited, although the development of Java tools should be coordinated with the education initiative.
There were strong opinions expressed that a survey wouldn't be necessary since many campuses are actively involved in creating multimedia educational materials. It was pointed out, however, that there is no coordination among these developers and that their creations weren't generally accessible. A goal of the initiative would be to set standards, possibly to identify common authoring tools, possibly to establish a library or regional libraries, possibly to coordinate the content of the materials. A broader goal is to give professors tools that make it easier to do things like use real-time data in the classroom.
Committee members were asked what UCAR's role should be. Some ideas expressed were to provide a Web site with COMET materials (de-construct the COMET modules and put the pieces on the Web), provide chat lines or e-mail discussion groups, create an on-line version of a standard curriculum.
The need for help with content is behind part of the users concerns. There are new concepts, new data sources, new research results and new emphases on and opportunities for interdisciplinary activities.
The UOP is proposing a needs-assessment because the UOP did not want to presume to know what is needed first. Can't "just do it" because we don't know what "it" is.

Discussion on Operating Systems

Discussion

Any strategy is contingent on the plans of the software developers: How do they plan to maintain current capabilities in the face of evolving platforms? What evolution in capabilities are they planning?
Unidata-supported applications need to be consistent with the broader network-centric world; we may find that the developers are reluctant to follow this evolution, however.
Unidata is a "utility" whose work of providing software and data must continue while it is engaged in developing new software.
Unidata and its developers are inter-reactive and interdependent.
OS/2 is a dying operating system; Unidata needs an exit strategy from this OS, but an alternative isn't clear. SSEC itself would prefer not to have to support it. However, it is more easily administered than any UNIX system, which is critical to some sites.
Historically, OSs are sunset only when new releases cease.
Resource question: at what point does it become cost effective to pull the plug on OS/2 and simply supply support for a basic UNIX system?
Before we exit OS/2, we need an entrance strategy to something else. too many institutions are relying on this technology.
The most immediate promise of Java is a simplified interface to UNIX administration. However, a Java-based replacement of current display and analysis capabilities will take much longer.
It is easier for universities to obtain funds for hardware when Unidata announces in advance that it will be necessary.
Perhaps the model of having a complete environment in every box is wrong; maybe we should strive for one ingester plus a bunch of browsers.
If Unidata is considering a sunset of OS/2, the community must be made aware of this because of the long lead time in the university funding cycles. It should also be a factor in the funding of the NSF equipment grants.
Sites' access to the Internet was a similar situation when the Policy Committee was considering IDD implementation
The UPC needs to examine its options; polling users will probably not be helpful in this.
The sunset of OS/2 should not be part of the new proposal; it should be completed before the proposal period begins.

Resolution 1:
The Policy Committee recommends that the UPC develop an exit plan for OS/2 that results in the sunset of OS/2 by June 1998. This plan will be considered by the Policy Committee at its next meeting. [passed unanimously]

Resolution 2:
The Policy Committee recommends that the UPC develop recommendations concerning operating systems (or substitutes for operating systems) for support by the Unidata Program and present them at the next Policy Committee meeting. [passed with one dissent]

Action 4:
There needs to be a newsletter article on UPC's examining options for the sunset of OS/2 as soon as possible.

Action 5:
Operating systems will be a topic on the agenda for the next Policy Committee meeting.

Unidata MRI Proposal

At the October 1996 meeting of the Policy Committee, it was suggested that Unidata consider joining a UCAR proposal for MMIA funds. On examining UCAR's existing plans for submitting a proposal, however, there was no clear role for Unidata. Fulker decided instead to focus on MRI (Major Research Instrumentation) funds. Unidata and UNAVCO investigated the possibility of placing GPS receivers at universities and polled the universities for interest. The deadline for submitting the proposal is short, however.

Discussion

It was the clear sense of the committee that the UPC should wait and apply next year.

Proposal to NSF for Future Funding

Copies of Fulker's draft project summary, proposed objectives, risk/advancement table, and software architecture diagram are in the notebook; a draft resource allocation table was distributed at the meeting.

Discussion

On the software architecture diagram:
- It represents the most coherent overview of Unidata seen recently.
- It doesn't address the evolution of packages; the diagram needs to show the capabilities of today and those of the future.
- The capability of interactivity doesn't come across.
On the risk/advancement table:
- Need to define risk somewhere: some endeavors represent high advancement or opportunity with no risk. Essentially, we're trying to fit multi-dimensional assessment onto a two-dimensional diagram.
General comments:
- UPC facing the problem of how to undertake development when support and development tasks are handled by the same people; invariably, development gets shorted by the more immediate demands of support. Implementing IDD required focusing on development at the expense of support. What's the model for Java development?
- GeoF is much bigger than IDD. Java development will require shifting resources. Some development may be handled by contractors and contributors, but UPC should consider building a resource bump into its budgets for development (estimate that the Java development will required $800k-$1.2 M) Don't want the resource bump to distort the Unidata Program, however. An alternative is to extend the time line.
- Maybe a more effective way to organize development would be to have a dedicated development group. It's a basic tenet of Unidata philosophy, however, that everyone needs to be in touch with users and involved in keeping their skill up to date.
- The proposal should articulate management philosophy. Specifically, UPC's model of keeping developers in touch with users, maintaining the flexibility to alter focus, and having a tight linkage between management and governance.
- UPC is facing a choice: Java or OS ports; can't do both. essentially UPC has chosen Java instead of "treading water" or going with NT.
- The proposal needs time lines.
- The UPC needs to map development with milestones and resources.
- The proposal should refer to the program's track record (a "trust us" paragraph that points to Unidata's experience with adapting to changing environments; failures should be acknowledged with a statement of what has been learned from them); this section should be linked to the management philosophy section.
- Since NSF has limited funds, perhaps Unidata should seek out opportunity funds (e.g., MRI, KDI funds) to fund Java development.