Policy Committee Meeting Summary:
May 29-30, 1997

Boulder, Colorado

Participants

Members	Representatives	UPC Staff
Otis Brown (Chair)	Richard Anthes (UCAR)	Sally Bates
Ken Crawford	David Fulker (UPC)	Steve Chiswell
Robert Fox	Robert Gall (NCAR/MMM)	Ben Domenico
Colleen Leary	Jewel Prendeville (NSF/ATM)	Jo Hansen
John Merrill	Mohan Ramamurthy (U. Ill./Users)	Linda Miller
Jim Moore	Other	Sandra Nilsson
Julie Winkler	Richard Chinman (UCAR/IITA)

Administrative Matters

The minutes of the January 1997 meeting were approved.
Jim Moore will be the Policy Committee Representative to the Users Committee.
John Merrill became the Policy Committee Chair at the end of this meeting.
Five notebooks containing the Policy Committee Resolution Log and the Unidata sites chart will be provided at the meetings; these materials will no longer be included in the notebooks mailed to committee members prior to meetings.
The next meetings of the Policy Committee will be:
30 September - 1 October, at NSF headquarters in Arlington, VA
5-6 February 1998, in Boulder, Colorado (as agreed via e-mail)
21-22 May 1998, in Boulder, Colorado

UOP Reorganization

UCAR President Rick Anthes briefed the committee on changes to UCAR's management, including the resignation of UOP director Bill Pennell and the hiring of Jack Fellows to fill Pennell's position. Kathryn Schmoll will be assuming the position of UCAR Vice President for Finance and Administration, replacing Bill Rawson, who retires in September. Both will assume their positions at the end of the summer.

Status Reports

Director's Report
Copies of Fulker's Director's Report were distributed at the meeting; Nilsson's budget report is in the notebook.

Discussion

UPC's current status and concerns appear to indicate a shift from operations to investment; this is a positive shift.
The NWS has other branches besides NCEP with software of interest to the Unidata community, viz., the Office of Hydrology.
There was considerable discussion about Unidata's future; the goals as outlined in the proposal to NSF are essentially those envisioned in the program's beginning (clothed in more modern terms and technologies); there was consensus that achieving these goals is important. At the same time, there was concern that the path to achieving the goals is not yet clear.
The UPC needs to clearly set its priorities and goals and organize its staff accordingly. (See the Operating Systems discussion below.)

Action 1:
The UPC will look into the software programs used in other branches of the NWS, particularly the Office of Hydrology.

Users Committee Report
The Users Committee plans for the summer workshop, including the draft program, are documented in the notebook. Ramamurthy reported that 54 participants had signed up. In other arenas, Ramamurthy noted that:

Increasing the resolution of the satellite images included in the IDD had caused a few users some problems, primarily due to platform-configuration problems. These were rectified by UPC support staff.
The Policy Committee's resolution concerning OS/2 caused quite a stir.
The Users Committee was pleased with UOP's PAGE proposal to NSF as a response to their concerns about the use of technology in the classroom.
Users are puzzled by the contractual constraint to wait 48 hours before putting radar data on the web, since vendors themselves are putting data on their web sites, sometimes within the hour.

Discussion

Nilsson contacted WSI re NIDS constraints; WSI is not interested in amending the terms of the contract at this time.

NOAA Report
Bob Fox spoke with Doug Sargeant; a summary of his conversation was distributed at the meeting. Fulker had also met with Sargeant prior to the Policy Committee meeting and reported that they discussed university/NWS interactions, among other topics. Sargeant believes that the NWS is willing to share software with universities but the mechanisms for doing so are complicated.

Discussion

There was discussion on whether lessons learned in the university could be transmitted back to NWS; the SOOs (plus having sites co-located with universities) provide one mechanism, but the recent budget cuts may jeopardize this since many SOOs may be moved into operational work.
There was concern about the progress of AWIPS deployment.

NSF Report

Jewel Prendeville reported on NSF for Bernard Grant, who was unable to attend due to his active status as a new parent. Copies of Prendeville's report were distributed at the meeting. Not on her slides were the following pieces of information:

The education effort within NSF/Geo is being led by a team composed of one representative from each division.
NSF/Geo already has many educational initiatives, but the new thrust is to reach out to elementary schools, not a usual audience for NSF/Geo.
For Unidata, the educational effort to watch is KDI: Knowledge and Distributed Intelligence, which is still being defined.

In addition to her report on the education efforts at NSF/Geo, she noted that:

The Unidata Proposal will be reviewed by a panel that will be convened in Boulder in September; panelists will be expected to provide written reviews.
The proposal for a new UCAR education initiative (named PAGE) is at NSF now; there will be eight reviewers; NSF expects a fairly rapid turn-around on the review process.
Bob Corell (NSF/Geo) believes that congressional cuts to research have bottomed out..

Operating Systems: Issues and Plans

Fulker outlined recent events on the issue of which operating systems the program should support. He noted that, in addition to the goals set in the proposal to NSF and the resolutions listed on the agenda for this meeting, the UPC was under pressure to support LINUX. Unfortunately, LINUX, while running on PCs, is still a UNIX system and therefore difficult to administer; further, there are many variants of LINUX, so it potentially would add to the support. burden. Therefore, the UPC recommends not supporting LINUX.

The question to the committee was: How will Unidata achieve platform independence? Fulker presented his view of Unidata's goals and a strategy for handling the transition (copies of his presentations were distributed at the meeting). His presentation was entitled "Toward Java and Platform Independence."

Discussion

There was complete consensus on Unidata's goal of achieving platform independence as soon as possible, and the Users Committee concurs with the program's goals as stated in the proposal.
Resolution 9701 caused turmoil within the community. Members of the Users Committee felt blind-sided by the resolution. Worse, they were left with the impression that the issue was one of Unidata support for large schools as opposed to small schools.
The community needs to be educated on why Unidata set platform-independence as a goal; this entails clearly articulating the pros and cons of Java (through a cost-benefit analysis) and addressing the platform-administration issues.
Handling the transition is the problem. It requires setting clear priorities, articulating an interim strategy for OS/2 sites; and deploying resources appropriately. The consensus needs to be community-wide, not just within the committees.
The pace of technological change in contrast with the slowness of university funding cycles makes it imperative to involve the universities as quickly as possible.
Unidata's user-level software is developed by partners; one support person works with each developer, since the UPC paradigm is to assign one support staff to one package. Any move to increase the number of supported packages quickly hits this barrier since there are no resources to add support staff. This paradigm may no longer be useful in the new environment of many more smaller software pieces. The UPC may need to redefine how it uses its support staff and expand how it uses "community support."
It is not clear who is to do what to achieve Geosif. Clearly, partnerships will be needed, but the close relationship between Unidata and one software developer may no longer be possible. The effort may need to entrain many developers, all of whom need to feel they have equal access to designing the interface.
The current clear distinction between Unidata-supported software and other software may become fuzzy. There is a potential for user-developed packages to out-pace Unidata-developed packages.
In planning for the new environment, flexibility is paramount.
If Geosif is to be achieved by 2003, a prototype should be ready in two years.
The SSEC will need to support OS/2 for the foreseeable future for its own customers.
The UPC is experimenting with a client-server implementation of McIDAS to handle the transition period. There were concerns about performance on PCs: the UPC needs to know precisely what systems users have and how they are used if this model is to be useful. The UPC envisions supporting parallel systems (client/server McIDAS and the evolving Geosif capabilities) for most of the next proposal period.

Resolution 1:

In view of

Policy Committee Resolutions 97.1 and 97.2,
the Users Committee Resolution to the UPC from its March 1997 meeting, and
the UPC conclusion that developing an effective plan responding to the Policy Committee resolution 97.1 is not feasible,

the Policy Committee endorses the principles expressed in Dave Fulker's presentation, "Toward Java and Platform Independence," namely:

The Unidata Program's long-term goal is platform independence
Target architecture is the Java virtual machine
The language of choice for new implementation is Java
The Unidata Program Center's roles are three:

Providing data management software
Entraining outside developers of user-level applications
Providing software development infrastructure.

Resolution 2:

The Policy Committee requests the UPC develop a transition plan from OS/2 and UNIX to a system involving a Java-based client-server implementation. The transition should include the following steps:

Development of a Java-based "image viewer" client that would serve as the GUI;
Adaptation of OS/2 and UNIX applications to function as servers to the image-viewer client.

Functions should be added to the client-side Java application as quickly as practical, eventually allowing it to replace platform-dependent systems. The transition plan will be developed by the UPC and discussed by the Users Committee, ATAC, and the Policy Committee; it will include a time line and a resource allocation table. The plan should include support for OS/2 and UNIX users until the capability of the Java application addresses most user needs.

SuomiNet

Developing a network of universities receiving GPS data will be the subject of a proposal to NSF next year for funding under the Major Research Instrumentation (MRI) program. With the permission of the Suomi family, the network, if it materializes, will be called SuomiNet.

Steve Chiswell presented the capabilities of GPS data and demonstrated how these data could enhance current models and be used in other disciplines such as engineering.

Discussion

UNAVCO GPS receiver packages can record temperature, relative humidity, and surface pressure in addition to parameters related to time and position; UNAVCO is hope to develop a slant-path algorithm to derive the gradients of precipitable water.
The cost of a GPS receiver is currently about $15,000 plus the mounting costs (which can be extremely expensive if monumenting is required). UNAVCO is exploring avenues to reduce the cost to about $3,000. The system is quite robust.
The terms of MRI funding require matching (30%) funds.
Unidata's role in SuomiNet would be to help sites employ the LDM/IDD for both gathering and distributing the data; data processing would be done at UNAVCO. (Data will be delivered to FSL as well as UNAVCO.) SuomiNet would demonstrated the IDD principle that any point on the net can be a data source.
There was discussion on whether or not to try controlling coverage by encouraging sites in some areas. However, multiple, geographically close sites can provide data on mesoscale boundaries or on microprocesses, so dispersed receivers are not necessarily the ideal.
The GPS data have a high teaching value; a good method for meeting university needs for mesoscale-resolution data sets.
Other (non-university) GPS sites are being planned; there are likely to be GPS networks implemented by departments of transportation and airports, for example. Can envision data flowing from all these in the more distant future.

Strategies for Distributing NEXRAD Data

Fulker reminded the Committee of his letter of April 1996 to the National Weather Service (in the notebook) commenting on the future of NIDS. University participation in NIDS is limited due to costs and to vendor policies. Sargeant had indicated that the NWS may be considering changing its arrangements with NIDS vendors since the service itself has some concerns about the current arrangement. Arriving a new arrangements, however, is difficult since there are several levels of bureaucracy that need to be involved.

Fulker outline possible Unidata courses of action:

Entraining the private sector direction, meeting vendors concerns through WITI (UCAR's commercial arm) on the assumption that WITI could apply Unidata software enabling vendors to create products with value-added components.
Working in parallel to the private sector.
Limiting Unidata's role to the academic sector where universities connect to NEXRAD via the Next Generation Internet and IDD; perhaps apply for USWRP or KDI funds to facilitate this.

Discussion

There are a total of only 2800 subscribers to NIDS (including special subscribers, government agencies, universities, private business, etc.) Unidata and the NWS together provide $70,000 as a subsidy for university access, which costs $200,000. This subsidy, however, does not bring the costs down enough for most universities. Universities reaction to the cost of NIDS is analogous to their response to the NPS data stream.
If Unidata establishes a floater site for the community, the community will want the raw data, not just the data products.
IITA will be establishing a NIDS data server and archive for UCAR, which could be accessed by Unidata universities; IITA is also talking to vendors about establishing floater sites to supply field programs with NIDS data in real-time.
Fulker would like to see a system that allows any user to select any radar; he assumes most users most of the time would select most-closely sited radar; users would pay proportionally by volume of data received. When the system matures, its network configure would allow a site to subscribe to as much data on a daily basis as desired.
The academia-only option listed by Fulker would be expensive. Universities need not only data but the infrastructure (software, knowledge) to use it. This situation is reminiscent of the early days of satellite images: most users used only the products from satellites; now more universities use images as data inputs.
Universities need to begin working with these data now so that the next generation of meteorologists understand how to use the data. The few universities that are pursuing access are finding that negotiations with the NWS and the vendors is a long and arduous process.
Oklahoma has fiber-optic lines to all the radars in the state and will be supplying Level II data to the OU, which will be incorporated in models. UniSys is allowing OU to redistribute the data in real-time to government agencies (at $25/month each) and K-12 schools (at $2/month each). The universitiy expects to eventually have 250 subscribers.

Action 2:
The Unidata users Committee needs to articulate the needs of the Unidata community for NEXRAD data streams, both currently and in the foreseeable future.

Strategies for Distributing NOAAport

The cost of FOS will increase with time as fewer and fewer people access it. What should be Unidata's plan for the transition? Fulker identified the following alternatives:

Maintain the status quo, with Unidata subscribing through Alden.
Explore the possibility of finding another vendor (but the origination fees are significant).
Become a direct subscriber to FOS (but Unidata could not function in full "operational" mode with 24-hour personnel).
Accelerate access to NOAAport.
Build a separate system.

Discussion

The current system with Alden has redundancies.
AWIPS software is available through the Forecast Systems Lab.
All FOS data are on NOAAport although some international data may be restricted.
The problems at Alden are unlikely to blow up quickly. UPC staff have remote access to Alden machines and can help troubleshoot when/if Alden's staff fail.
There appeared to be consensus that the UPC should maintain the status quo.

UOP PAGE Proposal

Sally Bates informed the committee of UCAR's response to the September 1996 Users Committee resolution on obtaining help from Unidata in using technology in the classroom. As a result of that resolution and a similar one from the October meeting of the AMS Heads and Chairs, Unidata and COMET began collaborating on a response. COMET has reorganized its Educational Resource Center into the Program for the Advancement of Geoscience Education (PAGE) program, now a new UOP program under the interim directorship of Mary Marlino. Marlino, with Unidata's Ben Domenico as co-principal investigator, has submitted a proposal to NSF to fund an assessment of what the university community needs and to develop a consensus on what the next steps should be. The proposal differs little from the concepts presented to the committee in January. As part of the process of developing the proposal, there is now a PAGE steering committee (a list of the committee's members was distributed). NSF has sent the proposal out for review and expects the decision on funding to be forthcoming.

Discussion

There was little discussion, although there appeared to be consensus on the appropriateness of UOP's decision to form PAGE.

NASA Cooperative Agreement Notices (CANs 1 & 2)

Copies of the agreement notices are in the notebook. Bob Fox was at the preproposal conference and made the following observations:

The announcements are a result of a long process of review that found problems with the contracts and design of the DAACs (Distributed Active Archive Centers). The CANs are in response and seek Earth Science Information Partners (ESIPs) to build a new system There are three levels of ESIPs: ESIPs 1 are the original DAAC contractors; ESIPs 2 are the research/university community, and ESIPs 3 are people in the private sector.
CAN 1 is directed at the ESIPs 2; it envisions having about $12 M available for grants; recipients will be expected to form a self-governing federation. Proposals will be evaluated on the science, the contribution to data processing technology, and the contribution to the federation.
The grants are for three years; there is not guarantee of subsequent refunding; there is not guarantee that the technologies resulting will remain proprietary to the grant recipients.
The "Unidata Model" was explicitly called out at the conference; there didn't appear to be any understanding of what the model represents.
Deadlines: June 11, letter of intent due; July 14, proposal due.

Fulker and Fox identified three possible options:

Ignore the CANS'
Submit a proposal that enabled access to Mission To Planet Earth data; or
Be receptive to being a subcontractor on someone else's proposal.

Discussion

Proposals need to have a central scientific theme and Unidata isn't a science-research group.
The DAACs will not be operational by launch time; NASA will need something like the IDD as a result.
Unidata has already been approached by the DODS (Distributed Ocean Data System) group at Rhode Island.
There appeared to be consensus that Unidata should be receptive to being a subcontractor on the DODS proposal.