Policy Committee Meeting Summary:
September 30 - October 1, 1997
NSF Headquarters, Arlington, VA
Participants
Members |
Representatives |
UPC Staff |
John Merrill (Chair) |
David Carlson (UCAR/ATD) |
Sally Bates |
Otis Brown |
Harry Edmon (U.Wash/ATAC) |
Steve Chiswell |
Ken Crawford |
David Fulker (UPC) |
Glenn Davis |
Robert Fox |
Bernard Grant (NSF/ATM) |
Ben Domenico |
Colleen Leary |
Clifford Jacobs (NSF/ATM) |
Jo Hansen |
James Moore |
Mohan Ramamurthy (U. Ill./Users) |
Linda Miller |
Charles Murphy |
Douglas Sargeant (NOAA) |
Sandra Nilsson |
Julie Winkler |
Other
|
|
Mary Glackin (NOAA) |
|
Paul Hershberg (NOAA) |
Administrative Matters
- The summary of the May 1997 meeting was approved.
- The next meetings of the Policy Committee will be:
February 5-6, 1998, Boulder, Colorado
May 28-29, 1998, location to be determined.
Status Reports
Director's Report
Copies of Fulker's Director's Report and presentation were
distributed at the meeting; a copy of Nilsson's budget report is
in the notebook. Linda Miller reported on the status of the
case-study effort; the Web-based versions are underutilized.
Discussion
- Unidata creates multiple "virtual" relay trees on
the IDD to decrease latencies. This involves separating the large
products into their own stream, which decreases the average
latency of the small products since they are no longer are held up
behind the large products). Large products are more likely to
require retransmission during periods of poor network performance,
and therefore tend to backlog other products behind them. When the
feed is split into separate streams, the smaller products can get
through even if the larger products, such as the MCIDAS satellite
images, do not. Most routers employ a round robin sharing
algorithm. Separate connections increase the allotted time for
the LDM for each cycle through the table.
- There was considerable discussion on the UPC's transition to
Java. The goal of machine independence was reiterated, but the
shape of the new development effort--scheduling development and
manage legacy software--remains poorly defined and somewhat
problematical.
- ATD, like Unidata, is planning to transition to Java for many
of the same reasons. Carlson suggested there may be synergies in
the UPC and ATD efforts.
- Planning may require developing a business model and
specifying who are the UPC's partners because UPC does not have
the resources to tackle this problem alone.
- The underutilization of the case studies may be a result of
their size--users need to have extensive disk space to hold the
data and few institutions have that much free space.
Users Committee Report
Ramamurthy summarized the Unidata/COMET summer workshop. There
were over 50 attendees, with 14 speakers and, although the Users
Committee has not had its formal debriefing, the exit interviews
indicate that it was a very successful workshop indeed. The
attempt to mix research and education topics, however, was not
altogether successful: attendees looking for in-depth instruction
on satellite meteorology left dissatisfied.
In a separate report, Murphy summarized his proposal that the
Users Committee form a working group to facilitate the assessment
of users needs and desires in the UPC planning for the Java
transition.
Discussion
- The Users Committee may want to consider selecting a topic for
its next workshop fairly quickly in order to submit the topic as a
possible ASP Colloquium. (The colloquia are set two years in
advance.)
- There was discussion of using the PAGE needs-assessment focus
groups as a mechanisms for obtaining input on community
needs.
- Murphy's concept of a working group fits in with the
domain-analysis process; working group members could test new
software as well. There is still a need for broader input,
however.
NOAA Report
Mary Glackin reported on the AWIPS implementation. She noted
that COMET would not receive a NOAAport receiver until July 1998
at the earliest. The formal commissioning of NOAAport is now
slated for 1998 with the weather service itself expecting to start
depending on NOAAport sometime during the year. As part of the
effort, NOAA is developing a method for handling communication
between external users and the National Weather Service. The
NOAAport stream will contain re-mapped satellite images and
model data (grids, no graphics) as well as observations; the
schedule for delivering larger grids (International Data Service)
has not yet been determined. The NWS is currently working to
merge radar with model data, so radar data will be part of the
NOAAport stream. NWS is exploring mechanisms for making more radar
data more widely available, which includes moving toward
non-proprietary open-access systems. How quickly this can be
achieved is unknown, but the current situation will change to some
extent when the NIDS contract expires. Lightning data on NOAAport
is the only restricted portion of the stream, but NOAA is
reserving the right to restrict other data in the future.
Discussion
- The National Weather Service wants to open up discussion with
the university community on the topic of data compression.
Compression, however, poses some problems to some users;
university users will want to sample areas of data, which can't be
done on compressed data.
- One role Unidata might play is to sample the NOAAport stream
for users with less powerful platforms.
NASA Report
NASA representative Hasler was unable to attend.
NSF Report
Jacobs reported on the current status of NSF's 1998 budget as it
wends its way through the budgeting process. He noted that the
Geosciences budget as submitted to Congress contains $34.37 M for
KDI activities. He reported that he had received 12 proposals for
equipment grants and funded 11 of them. He also reported on the
Integrated Spatial Information Systems (ISIS), which has as its
goal the achievement of full interoperability of geospatial
information.
The Unidata proposal will be reviewed after the current NCAR
review is completed. Since Unidata is already funded under the
UCAR Cooperative Agreement, there will be no special
"bridge" funding required to carry Unidata from the end
of its current proposal in May until the start of the new
UCAR Cooperative Agreement in October 1998.
Copies of his presentation were distributed at the meeting.
Discussion
- On ISIS, this is a very welcomed effort since, e.g., land-use
information is necessary to understanding climatologies; there
were questions whether the GIS community, which is much larger
than the atmospheric sciences, was really interested in
interoperability.
- NSF awarded four educational minigrants.
Unidata's Role in a NASA CAN Proposal
Fulker reported that Unidata has cooperated in the University
of Rhode Island Distributed Ocean Data System's response to a NASA
Cooperative Agreement Notice. A copy of Unidata's statement of
work is in the notebook. From Unidata's perspective, the DODS work
will allow Unidata to begin examining issues relating to
retrospective data. The DODS effort also promises an extension of
netCDF to allow links to URLs instead of filenames, which would be
useful to Unidata case-study effort.
Discussion
- With the DODS server, one can envision making data available
at various locations; location of data would be transparent to
user. The limitation is Internet bandwidth. This is an avenue for
developing access to archived data.
Java Transition
Steve Chiswell demonstrated Java-TRANS, a prototype Java-based
tool he created for supplying users with GEMPAK products. Fulker
explained that Chiswell's work is an example of one way to
approach the problem of dealing with legacy software.
Unidata has not yet reached a point at which it has identified
precisely how the transition to Java will be accomplished. The
staff is learning about object-oriented design and how to use
JAVA; pieces of the technology are evolving at different speeds;
Unidata has limited resources and expertise; and the problem is
highly complex. As a result, no products have been identified, no
time lines have been established.
Discussion
- The magnitude of the task facing Unidata was acknowledged.
There was concern expressed that identifying where to focus
efforts and resources might not be achievable in a truly
innovative fashion. (Current software is really almost 50 years
old; new models are needed.) There is a clear need for
visionaries, in computing science and in the academic Geosciences,
to help guide Unidata's transition.
- Unidata cannot do this alone; there need to be contributions
from the community. Building infrastructure (designing software
solutions that apply to range of tools) won't excite
participation; building prototype tools will. Vertically
integrated tools will serve as "path-finders": they
will help identify the key elements, will give something for
community members to experiment with, and may promote
contributions from others. There was a question whether there
were people in the community with the requisite skills to
contribute to the effort, although help may come from outside the
Geosciences community in the way of various tools that can be
adapted for Geosciences use.
- Identifying key layers (such as data compression or
navigation) is as necessary as building integrated prototypes; UPC
needs to do both.
- One possible start might be a tool for analyzing radar data,
built in cooperation with ATD.
- Unidata is not alone; other people and groups are at a similar
point. Need to combine resources and talents, but how do we
entrain these others?
- While platform independence is the long-term goal, Unidata has
considerable investment in its current software. This investment
should not be simply thrown away.
- Visionaries alone are not enough: everyone needs to be
involved in the design process since it is a community enterprise.
Sense of the Committee:
The committee agreed that UPC efforts toward Java need to include:
- a role for visionaries
- a role for "big thinking"
- vertical applications development on a rapid-prototyping cycle
- horizontal module development on software that will be applicable
aross a suite of tools
- identifying other key dimensions
The committee also agreed that the transition will require a multifaceted
approach and that there are uncertainties concerning the deployment of
resources.
The Scope of IDD
Paul Hershberg, Principal Scientist in the Office of Meteorology outlined
the need for cooperation between the NWS, the USWRP, and Unidata in an
effort to distribute NCEP model data. A copy of the draft proposal for this
is in the notebook. The data currently reside on an overloaded machine in
NOAA's Information Center. Due to heavy demand, the NIC cannot meet the
USWRP's requirements for timely access to these data. At the request of the
USWRP implementation working group, representatives developed a plan to meet
these data requirements through use of Unidata systems. The proposal
involves passing the data to an Office of Systems Operations (OSO) server,
expanding that OSO server from an FTP server to an LDM server, and
configuring the OSO LDM server to inject the data into the IDD. Costs pose
the limiting factor: this activity isn't in the NWS mission, so the NWS
would need to contract out the work involved. Given the current NWS budget
constraints, money for such a contract is not available.
Discussion
- The immediate problem of making the model data available might be solved
more simply by asking a single Unidata site to obtain the data via FTP and
then to inject the data into the IDD. A single point of access from the
university community might substantially lower the demand on the NIC
server. To ensure operation timeliness, however, the single source would
need to have priority access, which implies a formal understanding with
NOAA.
- University demands for data are growing faster than delivery
systems. We cannot expect the NWS to provide more data to universities
than it provides to its own forecast offices. While the need for a
formal mechanism that provides university access is clear, it
cannot be at NWS's expense.
- If this ends up costing Unidata money, it may become a small vs large
school issue; what is the priority here, NIDS or model data?
- Perhaps the issue is being addressed at the wrong level; need to
involve Steve Nelson and Rit Carbone.
Resolution 1 (passed unanimously):
The Policy Committee expresses its appreciation to NCEP for providing the
opportunity to distribute data to the Unidata community.
Resolution 2 (passed unanimously):
A. The Unidata Program Center will coordinate with one or two IDD sites
to have them obtain NCEP grids from the OSO servers and distribute the grids
to the community via the IDD. Unidata participants will be informed
when the grids are available via the IDD. Assessment of the impact of IDD
distribution on the OSO server would be useful.
B. The Unidata Program Center should continue to work with NCEP and the
USWRP to develop, maintain, and evolve a mechanism for distributing data
via the IDD.
SuomiNet
Steve Chiswell summarized recent efforts toward developing
SuomiNet--a proposed university-based network of high-accuracy GPS
receivers. A copy of the vision document is in the notebook.
Chiswell reported that a recent poll of the community indicated
active interest by XX universities. If there is sufficient
interest, a proposal will be submitted to NSF for funding the
network.
Discussion
- To date, there's been no attempt to involve sites outside of
the U.S.
- There may be a way to integrated data from geodetically
monumented sites, but this has not been investigated.
- UNAVCO currently gathers GPS data but they are lacking the
ancillary data (upper-air temperatures, pressure, ground
temperature) to calculated precipitable water.
- Need to clearly articulate the science in order to excite interest;
should also address issues of data latencies and data-sharing
policies clearly and overtly. The requirements of NSF's Major
Research Equipment grants program should be single driving force.
Unidata and NSF's KDI Initiative
The Unidata program is unique; it is also a knowledge network. It may be
worthwhile trying to model the program. Fulker outlined some of the
components of the model. Copies of his outline were distributed at the
meeting.
Discussion
- By model, Fulker was describing a method to produce a business model
with identified inputs and outputs.
- Process re-engineering principles are invoked when an entity is
broken; is there really a need to apply this to Unidata? Will it
really further understanding of Unidata or bring more money into the
program? This effort may benefit the program, however, by identifying
areas of needed improvement (e.g., data-flow optimization)
- A dynamic model may help explain how necessary it is to have an ethic
of sharing data.
- Meteorology has been transformed over the past 10 years: there are
new instruments, new software tools, a trend toward integrating
teaching and research, explorations of electronic textbooks and lab
manuals. Modeling Unidata might be a mechanism for clearly showing
how Unidata has facilitated this change.
- Need to devise mechanisms for measuring success.
- Modeling the Unidata Program is a separate
issue from the transition to Java. However, modeling the process of
the transition may be enlightening and aid the the effort to measure
success. It may also demonstrate the positives of Unidata's community
orientation/organization.
Brainstorming on Unidata and the Evolving Atmospheric Sciences
Undergraduate Curricula
Ben Domenico apprised the committee of the creation and status of UCAR's
new PAGE program. Copies of the PAGE overview are in the notebook.
Domenico noted that the locations of the focus groups have been established,
with the first occurring in Boulder on Oct. 17.
Discussion
- A formal mechanism for sharing data and information gleaned in the
focus groups is needed.
- Students may need to be involved in the focus-group efforts or in the
broader Page effort.
- The ability to deliver technology to the classroom is a big issue on
some campuses.
- Different pedagogies require different technologies; needs among
universities and among faculty within a single university will
diverse. Maybe PAGE might want to develop a matrix of university
needs, including components such as restructuring the curriculum as
well as technology issues.
Miscellaneous
Sally Bates reminded the committee that a program to recognize outstanding
Unidata participants had been suggested by the Users Committee and had
languished unaddressed by the Policy Committee.
Action 1:
A Policy Committee working group will be formed to address the Outstanding
Participation Awards program. The working group's task is to draft
an awards procedure before the next Policy Committee meeting.
Working group members are: Mohan Ramamurthy, John Merrill, Ben
Domenico, and Sally Bates.
Unidata Homepage
This page was Webified by Sally
Bates.
Questions or comments can be sent to <sally@unidata.ucar.edu>.
This page was updated on .