Policy Committee Meeting Summary: March 4-5, 1993

Boulder, Colorado

Participants

Members	Representatives	UPC Staff
Robert Fox (Chair)	Harriet Barker (UCAR)	Sally Bates
Otis Brown	William Bonner (UCAR)	Ben Domenico
Steven Mullen	David Fulker (UPC)	Linda Miller
John Nielsen	Robert Gall (NCAR/MMM)	Sandra Nilsson
Carlyle Wash	Clifford Jacobs (NSF)	Guest
Gabor Vali	Mohan Ramamurthy (Users Committee)	Dennis Walts (NOAA)
	John Snow (UCAR Board of Trustees)
	Timothy Spangler(COMET)

Administrative Matters

The summary of the December 1992 meeting was accepted without changes.
Next meeting: note dates and locations have been altered:
Thursday and Friday, June 17-18, in Washington, D.C.
Thursday and Friday, October 7-8, in Boulder, Colorado

Status Reports

Director's Report
Copies of Dave's report and transparencies were distributed at the meeting.

Discussion

Jacobs didn't see much chance of Unidata obtaining funds from the High-Performance Computing and Communications initiative.
There was a question whether the UPC should mount a booth and the AGU meetings. Some felt that the UPC should focus solely on meteorologists, others felt that trying it at least once might provide useful information. All agreed that mounting booths at the AMS was beneficial; it indicates that the program is healthy and allows the staff to introduce new capabilities.
There are still some sites not on the Internet. At worst, these sites may have to continue receiving data from Zephyr, although they may not be able to receive the Unidata/Wisconsin Channel or any other "experimental" data streams.

Action 1:
The UPC will draft a letter from the Policy Committee to sites not yet on the Internet. The purpose of the letter is to express the committee's concern about the quality of the site's access to Unidata Program services without an Internet connection and to ask whether there is any way the UPC can facilitate the site's gaining access.

Budget Report
Copies of Sandra Nilsson's transparencies were distributed.

Discussion
There were no questions.

External Programs Status
Copies of Miller's transparencies were distributed at the meeting. The FSL is offering partial support for UPC's effort to develop the Internet Data Distribution (IDD) system.

Discussion

FSL is interested in the IDD system as a way for them to get out of the special service arena; they currently have universities requesting special data sets of them. Also, they want use of the LDM to convey data to weather offices. There was some concern that there may be a mismatch in expectations between Unidata and the FSL.
Miller was commended for instituting the email list on NWS data changes; it's been very helpful to the community.
No one knows how the upcoming NSW data-stream changes will affect sites. UPC needs to know:
- whether format will change and if so, how.
- whether and how Zephyr will change
- whether the channel will be overrun
- what data rate will by required of the receiving platforms

Action 2:
The UPC will assess the stiuation and inform the Policy Committee via email. The UPC will also notify users of the impending changes as soon as possible.

NSF Report
Copies of Jacobs' transparencies were distributed at the meeting.

Discussion

NSF will be moving its headquarters this year.
The Internet Data Distribution white paper is being circulated around the NSF.
Jacobs is interested in familiarizing people with the Unidata Program as a model for how various communities can focus their efforts to solve particular problems and as a paradigm for how a community can leverage funds. HRE people particularly interested in the concept.
Funding for the 1994 Users Workshop has been cut by NSF from the Unidata budget; consequently, the Users Committee will need to prepare a separate proposal as was done for the 1991 workshop.
While NSF funding has increased, these increases have been earmarked for the strategic research initiatives.
Concerning equipment grants, Jacobs is concerned that small schools aren't applying and is interested in exploring mechanisms for helping them apply. Some suggested that the UPC offer to help small schools write proposals. Perhaps UPC should identify a specific UNIX configuration to recommend to small schools, which would make it easier for them to decide what and how much they need. Others suggested that NSF's requirement for matching funds from the institution discouraged many of the smaller schools. Jacobs indicated that NSF was flexible on this and that Unidata might pass this information on to the smaller schools.

Users Committee Report
Ramamurthy reported that the Users Committee has had several email discussions on the 94 Users Workshop (copies of the email exchanges are in the notebook). The committee had counted on receiving funds from the UPC, but now realizes that those funds were not included in Unidata's award.

The user survey ranked software integration as having relatively low priority, with compatability in data formats being ranked highest.

There will be a regional workshop this summer hosted by SUNY-Brockport the week of August 15. Brockport is requesting participation by Bruehl, Yoksas, and, to help with equipment installation, Schmidt.

Discussion

NSF will need about 6 months to consider workshop proposal. ASP and COMET have both indicated some willingness to contribute support for the workshop. One proposal could serve all three potential sponsors: NSF, COMET, and NCAR/ASP. Total costs are assumed to be in the $80 ,000 to 100,000 range.
There was discussion about the focus of the workshop: Users Committee sees the NWS modernization effort as the major focus, but should K-12 educators be involved? There is concern that involving them might be difficult and dilute the focus of the workshop.
Three committee members are due to retire; nominations for replacements are needed.
There was discussion about the advisability of updating the Huffman matrix; might raise expectations of developing a monolithic package; this could divert the UPC over the short term. However, a "requirements" document, based on users opinions, might be a useful guide in Unidata's development efforts.
There is the perception that Unidata has benefited the small schools more than the large ones; the objective of the program is to bring up mean state of the community, "to diffuse the capabilities of the few to the many." To do this, need to incorporate the ideas of the few (the "quantum leapers"), so the program needs input from the large schools.
Should the UPC consider supporting HP because of the AWIPS decision? The question was deferred to the AWIPS briefing discussion.

Action 3: The Users Committee should survey the user community before the next workshop. A goal of the survey is to update the statement of application requirements.

Internet Data Distribution Update
Copies of Domenico's slides were distributed at the meeting.

Discussion

What incentive is there for a university to become a relay site? The UPC envisioned the data distribution functioning like UseNet, with sites contributing machine time and expertise in return for receiving the benefit of receiving data on line. Some members of the committee questioned this, particularly in view of the fact that the relay site would receive all the complaints from downstream sites when the system failed. UseNet, afterall, is not time-dependant, nor are there faculty building classroom lectures around the service. If the UseNet model is followed, may need to make receiving data contingent on being willing to act as a relay site.
Some members suggested that centralized administration of the Internet Data Distribution system would be preferable since it would allow centralized control over authorizations and the LDM product tables. Administering a centralized system would soak up a lot of UPC resources, however. Does the UPC want to become a data-management facility or remain in technology development? Perhaps UPC should explore the possibility of finding another institution to manage the system once it's installed, e.g., SCD (which is already an around-the-clock operation). What happens when a university wants to distribute data? How do the distribution agreements affect UPC's ability to distribute data?
Some members suggested that the UPC consider purchasing the relay-node hardware. This would make the system easier to manage. It would also ensure a homogeneous system, circumvent access issues, and make it easier to separate ingest and decoding functions. It would result, however, in trading off a deep distribution sytem (with few top-level nodes) for a shallow one (more top-level nodes, fewer layers). The UPC might be able to increment system slowly, increasing reliability by adding machines. If the UPC were to purchase all the relay machines, this would require writing another proposal to NSF. However, these questions will not interfere with the UPC continuing to test the concept. The UPC is planning to identify test sites and will ask them what they're willing to commit to. Have 6-12 months before the UPC needs to identify official relay sites.
Choosing a model for the system is impossible at the moment, since the UPC doesn't have experience with high-volume data and so don't know how this will affect the system fans.

Action 4:
The UPC should try developing a model of the Internet Data Dsitribution system in the near future and distribute it to the Policy Committee. The questions to be addressed are: centralized vs. distributed administration, homogeneity of platforms; broad vs. shallow distribution, and the size and quality of the burden on relay sites. Many of these will be addressed in the testing process, but some may require a statement of philosophy.

YNOT Update

Fulker reported that the UPC has identified four test sites, and is now proceeding with the tests.

Discussion

Testing will likely take 9-12 months.
MDA is concerned by YNOTs slow acceptance by universities; has indicated a willingness to work with UPC on the testing process.

COMET Case Studies

Spangler reported that COMET is developing case studies for its courses. In the course of this, they found that the FSL workstations are not sufficient for all users. Consequently, COMET worked with UPC's Bruehl to develop a bridge to use GEMPAK display capabilities in the case studies. If there is sufficient interest, COMET could publish them on DCs and would be willing to distribute them to the universities. The case studies vary in size from 300 MB to 2 gB.

Discussion

Images are in FSL workstation form and are remapped, so there were questions about how useful the case studies would be to universities.
COMET is not set up to support universities, so Unidata would have to provide the support. What would this entail?
Preparing useful case studies requires enormous resources and usually entails specifiying a toolset for their use. But there's not a common toolset within the Unidata community. And if one tool is chosen (GEMPAK), what about the sites that don't use that tool? The studies might need to be altered to be ingestable by the LDM: who should do this? Who would undertake other changes that might be required? UPC's role is to provide tools for the community to use in doing meteorology, not to do meteorology itself.
To be useful, case studies require an education component, which is not in the Unidata mission; but they do come with documentation prepared by COMET.

Action 5:
A group of COMET and UPC people should investigate the difficulty and the resource requirements involved in making the COMET case studies accessible to the Unidata toolset. A report on this should be on the agenda for the next meeting.

NEXRAD Updata

Fulker reported that there are three possible scenarios for distributing NIDS data:

NIDS handles the distribution
Vendor distributes the data to everyone on the Internet
Vendors distributes the data to one or two relay sites who distribute to everyone else.

The choice of scenarios will depend on cost.

Discussion

Several important questions need to be addressed: What happens when a site wants to switch radars? What happens when everyone wants to look at the data from one radar? Following the floater paradigm might be a solution.
The UPC is assuming that sites will be able to display the data using WXP and McIDAS.

AWIPS Update

Dennis Walts of NOAA gave the committee a thorough and enlightening presentation on current status of AWIPS. Copies of his transparencies were distributed at the meeting.

Discussion

Developers in the government will be provided with Open Developers Workbench; applications vendors will be given PROFS to convert to the AWIPS platform with specified functionality.
On-site expertise (at the weather offices) will be provided by the contractor, although there will be a new Weather Service position: Electronic Systems Analyst.
Communications will be .25 and TCP/IP externally, but Walts doesn't know what will be used within an office.
NOAA building a satellite broadcast network; universities may want to invest in NOAA receiver to receive broadcast directly: the antennas will be 1.4M and 1.8m and relatively inexpensive.
The government will be splitting GOES images into two separate streams and will broadcast remapped, 8-bit data only. Full disks, east and west, won't be transmitted.
Government is leasing the system from developers, but owns the design of the system. The data are considered government owned and public. In the three to five years before full deployment the technology will advance. Consequently, the hardware is not fixed: the contractors are free to move to newer platforms--the NWS is not committed to HPs. Interfaces to the outside world will occur at Weather Forecast Offices and through a local data acquisition and distribution system.
AWIPS raises lots of questions for Unidata: licensing? software applications? Can/should the program duplicate what is essentially five T1 lines-worth of data with the Internet Data Distribution system? Will there be a workable interface between AWIPS and the Unidata system? What is the role of the Unidata/Wisconsin Channel? Committee members believed Unidata would have the Internet system fully deployed well before AWIPS is deployed. Since no one can predict where computer and network development will go, members believed that there's a good chance the AWIPS system will be out-of-date before it is deployed. Unidata should continue to help push the science.
AWIPS is clearly designed as a forecasting tool, not a research tool. University interests would not be served by the remapped images. Universities will want to analyze full satellite images. Furthermore, smart observing systems now being planned will drive a need for new analysis and display tools. Unidata needs to consider what it can use from AWIPS and what it needs to develop on its own.

Action 6:
Further discussion of AWIPS should be on the agenda for the next meeting, and probably for all the Policy Committee meetings over the next few years.

Other Discussions

The committee discussed the Domenico-Brown proposal to appoint "coordinators" to the Advanced Technical Advisory Committee (ATAC) rather than fixed members. There was some questions whether this would work, since the members might not feel ownership of the committee. While liaisons to groups with similar technical problems would be good, ATAC's highest priority is to help Unidata with the technical issues of the Internet Data Distribution system.

Action 7:
A report on the ATAC should be on the agenda for the next meeting. This report should include a list of current technical issues. The ATAC can hold an initial organizational meeting.

Several universities are interested in receiving ETA model data and MAPS analyses and have agreed to FTP these data amongst themselves; however, FSL is willing only to provide these data to UPC and NCAR, not to the universities directly. UPC is not in the position to distribute data at this time and there were questions whether it would be allowed to in any case given the distribution agreements currently in place. The committee suggested that the universities arrange to get these data from NCAR.
There was a demonstration of the Gopher service and the new Unidata Gopher. The UPC now considers the Gopher/WAIS accessible email archive to be its first level of user support.

List of Action Items

Preliminary draft done.

Action 2:
The UPC will assess the (FSL) situation and inform the Policy Committee via email. The UPC will also notify users of the impending changes as soon as possible.

Done.

Action 3: The Users Committee should survey the user community before the next workshop. A goal of the survey is to update the statement of application requirements.

The Users Committee is in the process of designing the survey.

The IDD management plan is in Section A of the notebook; a presentation on IDD is on the agenda; and the UPC is engaged in preliminary discussions with BBN on researching the network management issues.

COMET/UPC meetings have been held and this topic is on the agenda.

Action 6:
Further discussion of AWIPS should be on the agenda for the next meeting, and probably for all the Policy Committee meetings over the next few years.

Done.

Action 7:
A report on the ATAC should be on the agenda for the next meeting. This report should include a list of current technical issues. The ATAC can hold an initial organizational meeting.

Done.

Index

Unidata Homepage

This page was Webified by Jennifer Philion.
Questions or comments can be sent to <support@unidata.ucar.edu>.

This page was updated on .