HDF, EOSDIS, NASA ESE Data Standards

1
HDF, EOSDIS, NASA ESE Data Standards

• Richard Ullman

2
Agenda

• ESDIS Status wrt HDF
• EOSDIS (American Customer Satisfaction Index)
• NASA Earth Science Standards Endorsement Process

3
ESDIS Status

• Launch of Aura (July 25) marks end of development
  phase of the EOSDIS Core System (ECS).
• System is now in maintenance. Capability
  refinements are under the Synergy program.
• Data enters are now running Synergy 3 release.
  Will be transitioning to Synergy 4 over the
  next six months.
• Maintenance of HDF for EOS includes two
  components
• Support of NCSAs HDF group through a
  cooperative agreement.
• Support of HDF-EOS through ECS maintenance
  contract
• Other ESDIS project sponsored HDF-related work
  will be phased out near the end of calendar year
  2004.
• http//hdfeos.gsfc.nasa.gov website updates
• SESDA hdf data usability task
• Coordination, outreach and test bed development
  for HDF integration through CEOS, OGC, ISO
  organizations.

4
HDF-EOS

• A profile, convention, convenience API, etc for
  NASAs Earth Observation System standard data
  products.
• Defines structures for Point, Swath, Grid
  (Atmospheric Profile, Zonal Table)
• Defines specific location for product metadata
• ODL encoded metadata compliant with FGDC content
  standards.
• Maintained by a by L3-Communications under
  subcontract to Raytheons ECS Maintenance and
  Development contract.
• Next release expected Dec. 2004
• HDF5-1.6.3
• SZIP 1.2
• New inquiry functions
• CEA (Cylindrical Equal Area grid projection
• Improved performance in read/write functions 

5
HDF in NASA Earth Remote Sensing

• HDF-EOS is format for EOS Standard Products
• Landsat 7 (ETM)
• Terra (CERES, MISR, MODIS, ASTER, MOPITT)
• Meteor-3M (SAGE III)
• Aqua (AIRS, AMSU-A, AMSR-E, CERES, MODIS)
• Aura(MLS, TES, HIRDLS, OMI
• HDF is used by other EOS missions
• OrbView 2 (SeaWIFS)
• TRMM (CERES, VIRS, TMI, PR)
• Quickscat (SeaWinds)
• EO-1 (Hyperion, ALI)
• ICESat (GLAS)
• Calypso
• Over 3 petabytes of EOSDIS archived data

6
HDF-EOS Lessons

• Definition of a set of data structures as a
  profile is not sufficient to guarantee
  interoperability.
• Also need definition of content, especially
  metadata - this is increasingly difficult the
  wider the disciplines covered.
• See AURA DSWG standards and NetCDF CF as
  examples.
• Also need conformance measures - no spec is so
  clear that it cannot be misinterpreted.
• Even during life of mission, there must be
  allowance for technology refresh.
• Technology advances affect user expectations.
• Well understood concept for hardware -
  traditionally less recognized for science
  software and data products.
• See OAIS

7
Discussion topics today

• Ask the experts
• A growing number of software products depend upon
  the HDF libraries. Are there suggestions for how
  to better coordinate HDF library releases.
• Questions from participants.
• HDF-GEO?
• Last workshop there was strong opinion expressed
  that there should be some kind of bridge among
  HDF geographic and geophysical profiles.
• Can we develop a better sense of what such and
  HDF-GEO might be?
• Is this the list? HDF-EOS, NetCDF API, HDF-NPOESS
• What are reasonable expectations for this effort?

8
From ESDSWG meeting last week Why Use a
Standard?

• Good documentation
• Other projects have reviewed it and found it
  useful
• Reusable software sometimes available
• Potential users can see that standard and
  software works
• Not management pressure or peer pressure just
  more practical

9
2004 EOSDIS Satisfaction Survey
10
2004 EOSDIS Satisfaction Survey

• A measure of customer satisfaction
• ESISS and ESSAAC have recommended that NASA focus
  on measuring the impact of our systems and
  services rather than just the output
• In 2004, NASA used a comprehensive survey to
  determine the American Customer Satisfaction
  Index (ACSI) for EOSDIS products and services.
• ACSI provides a normalized measure of customer
  satisfaction that allows benchmarking against
  similar companies and industries.
• 2004 survey results show that customer
  satisfaction with EOSDIS compares very favorably
  with both industry and other government agencies.

11
Snapshot of the American Customer Satisfaction
Index (ACSI)

• The 1 national indicator of customer
  satisfaction today
• Compiled by the National Quality Research
  Institute at the University of Michigan using
  methodology licensed from the Claes Fornell
  International (CFI) Group
• Measures 40 industries and 200 organizations
  covering 75 of the U.S. Economy
• Over 70 U.S. Federal Government agencies have
  used ACSI to measure more than 120
  programs/services
• CFIs Advanced methodology quantifiably measures
  and links satisfaction levels to performance and
  prioritizes actions for improvement

12
Survey Background

• EOSDIS survey was performed by CFI Group through
  a contract with the Federal Consulting Group
  (Department of Treasury).
• Survey questions developed by the DAAC User
  Services Working Group were tailored to fit the
  CFI methodology
• ESDIS provided the CFI Group with 33,251 email
  addresses from users who had used NASA/EOSDIS
  products
• CFI sent invitations to participate in an online
  survey to 9,999 randomly selected users
• 1,056 responses were completed
• 1,016 surveys were used in the analysis (250
  responses were needed for statistically
  meaningful response).

13
EOSDIS Results

• The Customer Satisfaction Index for NASA EOSDIS
  is
• The Customer Satisfaction Index score is derived
  from customer responses to three questions in the
  survey
• How satisfied are you overall with the products
  and services provided by the Data Center (79)?
• To what extent have the data, products and
  services provided by the Data Center fallen short
  of or exceeded your expectations (73)?
• How well does the Data Center compare with an
  ideal provider of scientific data, products and
  services (71)?
• This score is four points higher than the 2003
  American Customer Satisfaction Index for the
  Federal Government overall (71).
• The confidence interval for ACSI is /-1.1 for
  the aggregate at the 95 confidence level.

75
NASA EOSDIS Aggregate Segment
14
Score ComparisonCurrent Location
74
ACSI
76
88
Customer Support
82
85
Delivery
83
USA (n478)
72
Product Selection
Outside
and Order
the USA (n577)
73
69
Product Search
71
67
Product Quality
69
34
Complaints
31
15
Customer Support - Score 84, Impact 1.0
Customer Support
84
CFI considers EOSDIS to be World Class in the
area of customer support.
87
Professionalism
Technical knowledge
85
Accuracy of information provided
85
Helpfulness in selecting/finding data or
84
products
Helpfulness in correcting a problem
83
82
Timeliness of response
16
Product Quality - Score 68, Impact 0.9
17
Analysis of Results

• Product quality is the lowest scoring component
  (68), and has a relatively high impact (0.9).
• All attributes in this area received similar
  ratings
• At 84 customer support scores well, and is also
  high impact (1.0).
• There is a significant difference in customer
  support ratings given by customers within the
  U.S. (88) compared to those outside the U.S.
  (82).
• The components product search, product selection
  and order are highly correlated.
• Recent customers are more satisfied, but are also
  reporting more problems.
• Percentage of customer complaints is fairly high
  (32) when compared to the federal government
  overall (12).
• Customers may not be calling to complain about a
  problem, but rather to seek assistance in solving
  the problem.
• 90 of respondents who answered the customer
  complaint questions gave user services complaint
  handling a rating of 6 or above.

18
CFIs Recommendations for Improving ACSI

• Focus on Product Quality
• Review the type of data product documentation
  available with each product. Work to improve the
  clarity and thoroughness of the documentation.
• Assess the various data formats and work to
  improve the usability of each.
• Offer a wider variety of data formats.
• Review the Product Search and Product Selection
  and Order scores to determine how best to help
  customers find the data they need
• Due to high correlation, improvements in one area
  will likely result in improvements in the other.
• Simplify the search process make data products
  more apparent.
• Improve data product descriptions.

19
Product Format Ease of Use Comparison
HDF-EOS HDF Geo-TIFF Binary ASCII
Valid Responses 270 190 61 53 44
Mean Valid Score 6.76 7.20 7.48 7.02 7.30
Median Valid Score 7 8 8 7 8
Standard Deviation 2.47 2.34 2.03 2.76 2.54
95 Confidence Interval 0.29 0.33 0.51 0.74 0.75
of Users Assigning 8 or More 46.7 52.6 55.8 49.0 63.7
The relatively low scoring of HDF-EOS was supported by users free text comments. The relatively low scoring of HDF-EOS was supported by users free text comments. The relatively low scoring of HDF-EOS was supported by users free text comments. The relatively low scoring of HDF-EOS was supported by users free text comments. The relatively low scoring of HDF-EOS was supported by users free text comments. The relatively low scoring of HDF-EOS was supported by users free text comments.
20
NASAs Earth Science Data SystemsStandards
Process
21
Insights

• Interoperability does not require homogeneous
  systems, but rather coordination at the
  interfaces.
• Management can judge success based upon program
  goals rather than dictate solutions.
• example degree of interoperability rather than
  use of particular data format.
• Communities of practice have solutions.
• Published practices that demonstrate benefit can
  grow
• successful practice in specific community
• broader community adoption
• community-recognized standards

22
The ESDSWG Standards Process

• Modeled on Internet Engineering Task Force RFC
  process and tailored to meet NASAs
  circumstances. The standards process provides
• Registers community practice for NASA
• NASA Earth science data management can rely on
  standards to achieve highest priority
  interoperability
• Encourages consensus within communities
• Science investigators are assured that standards
  contribute to science success in their
  discipline.
• Grows use of common practices among related
  activities
• Discipline communities benefit from the expertise
  gained by others
• Documents data systems practices for use by
  external communities.
• Lowers barriers to entry and use of NASA data.

23
Standards Process Group Strategy

• Adopt standards at the interfaces, appropriate to
  given science and drawn from successful practice.
• Find specifications with a potentially wide
  appeal
• Draw attention to a much broader audience
• Monitor use, promote what works well
• Result Accelerate the evolution and adoption
• Preferred source of RFC is community nomination.
• Possible to direct creation of RFC in response to
  identified needs.
• Consequence of endorsement
• Future NASA data systems component proposals will
  be judged partly on how well they interoperate
  using community-identified practices or else
  justify why departure from community has greater
  benefit.

24
Three Step Standards Process
25
SPG Review
26
Whats in the works

• DAP 2 standard used by many in the
  oceanographic community basis for the DODS and
  OpenDAP servers. -- submitted in June as a
  Community Standard
• Request For Comments on implementation
  experience distributed October 1, comments due
  November 12.
• Precipitation Community discussing potential
  science content standards being used to define
  level 2 level 3 data
• Self identified group of precipitation scientists
  have identified need and are proposing a draft.
  Are discussing at IPWG in Monterey.
• The community is establishing de facto standards
  in this area and that is the best way to deal
  with this.
• FGDC Vegetation Index standard discussing with
  potential community members

27
Ideas from the last ES-DSWG

• GCMD DIF
• GeoTIFF
• NetCDF CF
• OGC suite

28
Community Leadership

• Strong proposals will have
• Leadership to support and use standard
• Potential for impact
• Potential for approval
• Simple standard is better
• Potential for spillover to other communities
• Successful RFCs will have
• At least two implementers
• Demonstrated operational benefit
• Leadership in generating the RFC
• Community willing/able to review

29
SPG Contacts

• Earth Science Data Systems Standards Process
  Group
• http//spg.gsfc.nasa.gov/spg
• Chairs SPG
• Richard Ullman richard.ullman_at_nasa.gov
• Ming-Hsiang Tsou mtsou_at_mail.sdsu.edu