My perception of the fundamental problem

1
My perception of thefundamental problem

• Data are not sexy
• PI easily signs up to need to deliver
• But the published paper takes priority
• and then the next proposal takes priority
• Data managers are viewed as geeks
• who can only cajole, not enforce, data delivery

2
My perception of fundamental questions

• How can we raise the profile of data management?
• How can we persuade scientists to take it really
  seriously?
• NB carrot not stick e.g.
• Funding specifically for data management
• Show PIs how it can help them

3
Background - some of theGEOTRACES recommendations

• Appoint a Data Liaison Officer (DLO) at the IPO
• Appoint a Data Specialist on each Project to be
  responsible for data
• Agree time scales for data submission and release
• Use CCDHO for CTD and bottle data
• Involve data management professionals in all
  GEOTRACES data activities from the start

4
DLO responsibilities

• Maintain a list of IMBER projects
• Keep track of project metadata
• Maintain a catalogue of actual and expected data
  sets (DIF) or equivalent discovery metadata
  records (ISO19115 standard)
• Ensure that standardized parameter descriptions
  are adopted (e.g. the BODC parameter usage
  vocabulary)
• Ensure that protocols for naming of cruise,
  station positions, etc., adhere to a rule system
  developed by the DMC
• Interact with DAC(s) to coordinate their
  activities and interactions with PIs
• In particular, ensure timely delivery of metadata
  and actual data to the DAC(s)
• Contribute to (and possibly maintain) the project
  web-site

The DLO will be an ex officio member of the DMC,
and will report to the Director of the IPO and to
the DMC
5
Data Specialist

• Ensure that suitable log sheets have been
  provided for all activities
• Assist and support scientists in preparation of
  metadata
• Maintain regular checks that all logs are being
  correctly completed
• Assemble all metadata from a section or process
  study
• Assist with preparation of data files, ensuring
  that all necessary parameters are included
• Evaluate the quality of data, either by personal
  expertise or by discussion with PIs, and help to
  document quality and missing or suspect data
• Facilitate assembly of shipboard data sets and
  data integrity checks

6
Data Management for IMBER

• IMBER is a relatively young project
• But has taken early initiatives on data
  management
• Sophie Beauvais appointed as Data Liaison Officer
  (DLO) in May 2006
• Data Management Committee (DMC) appointed
  recently
• DMC discussions by email/blog? Getting off to a
  slow start - need to meet face to face

7
Data Management Committee

• Observationalists
• Raymond Pollard, Jay Cullen
• Modellers (data users)
• Wilco Hazeleger, Reiner Schlitzer
• Data specialists
• Todd OBrien, Gwen Moncoiffe, Toru Suzuki

Balanced to improve communication and mutual
understanding
8
Data v metadata

• IMBER manages no projects so strictly owns no
  data
• IMBER can encourage but not enforce improved
  standards for both metadata and data
• Should DMC emphasize the minimum (basic metadata)
  or the maximum (streamlined access to data?)
• Answer both

9
Metadata

• One view says it is hard enough to get decent
  metadata, so if we can achieve that it will be a
  major step forward
• Cruise Summary Report (CSR) - can we persuade all
  nations/labs to complete them (USA included)?
• Create DIFs at GCMD

10
Data

• Another view is that we must aim for accessible,
  high quality end data, arguing that nothing is
  more irritating than going through a dozen
  metadata links only to end up at data - contact
  PI - and of course the PI never answers
• Possible ways to achieve this
• Seamless access to widely distributed data
• Relatively small number of specialist data
  centres
• Both cost serious money

11
Specialist data centres(by which I mean,
specializing in a particular type of data)

• CCHDO (CLIVAR and Carbon Hydrographic Data
  Office), at Scripps, was WHPO (WOCE Hydrographic
  Programme Office)
• COPEPOD, the global plankton database
• Are these the way forward?

12
Why advocate specialist data centres?

• To gather in data from individuals to central
  archives
• To have the ability to quality check a particular
  kind of data
• To have specialist experience to help an
  individual PI

Useless without long term funding
13
Improving communication

• A major goal must be to get scientists and data
  managers talking to each other.
• Carrot, not stick. How can data specialists HELP
  scientists
• Backup PIs data at early stage - security
• Help with calibration
• Help with validation
• Long term archive
• Answer requests for data

14
Summary - points for discussion

• Raise profile of data management
• Improve communication (meeting, blogs)
• Carrots (funding, recognition, support)
• Specialist data centres (cf WOCE)
• Metadata standards (CSR, DIF)
• Adequate manpower