U.S. GLOBEC Scientific Steering Committee Coordinating Office Division of Environmental Studies University of California Davis, CA 95616-8576 Phone: 916-752-4163 FAX: 916-752-3350 Email: T.POWELL (Omnet) Email: firstname.lastname@example.org (Internet)
Additional copies of this report may be obtained from the above address
Precedent and perception have resulted in a disparity of data collection, storage, and archival methods. This makes the exchange of data difficult and may suppress dissemination of data. The U.S. GLOBEC Scientific Steering Committee seeks to enhance the value of data collected within the U.S. GLOBEC program by providing a set of guidelines for the collection, storage, and archival of these data sets.
The policy detailed below applies to all U.S. GLOBEC investigators. Field data, retrospective data sets, and numerical experiments must all be included in the U.S. GLOBEC database.
The second and seventh USGCRP policy statements address the need for exchange of data between researchers. A period of exclusive use is permitted, though the data should be made available when they become widely useful. The annex of the USGCRP expands upon this policy with the statement
"In the past, some Principal Investigators have retained data for indefinite periods, and this has inhibited their widespread use. This practice should be eliminated through active consideration of the tradeoffs between widespread distribution of data sets and the need to assure data quality and validity. The guiding principle is that as soon as data might be useful to other researchers the data should be released, along with documentation which can be used by the other researchers to judge data quality and potential usefulness."This clearly limits periods of restricted access to the time during which data is not generally useful. There is no provision for granting a period of exclusive use to provide the Principal Investigator with an opportunity to delay exchange until papers describing the data have been published.
Statements 4, 5, and 6 address the need to provide easy access. Statement 3 identifies the need to designate an archive for all relevant data. The policy must prevent the loss of important data sets. The annex notes that many data sets, especially biological, have no archive.
This document primarily addresses data collected during ship based field experiments. Extrapolation to other types of studies is expected. The policy statements anticipate that data will be organized about cruises; when measurements are not made aboard ship, the data should be organized about a period of time during which a sensible unit of data is collected. For instance, telemetered time-series data might be organized by month and near-shore benthic data might be naturally organized on a seasonal schedule. Investigators tend to organize their data into periods which are natural for their purposes. If a particular time period is appropriate for your study, then use this period for organizing and submitting data to the U.S. GLOBEC Data Management Office (DMO) while keeping in mind the need for timely submission. Investigators conducting retrospective studies should also recognize that the data organized for their purposes should in general be submitted to the DMO. Data culled from generally unavailable or difficult to access archives are of great value to the community. Model results which would be useful to the interpretation of field data or comparison with later model studies should also be included. Potential candidates for submission include annual or seasonal fields of flow, temperature, and salinity, Reynolds stresses, particle trajectories, initial/boundary conditions, and surface fluxes.
All principal investigators are required to submit plans for the collection of data prior to execution of their sampling program. In general, these plans are expected to be similar to the information provided in proposals submitted prior to funding. The purpose of this requirement is to provide a common resource for the participating scientists to evaluate the suitability of the expected data set for achieving their scientific objectives. A single description of the expected data sets, a "data plan", will be derived by the Data Management Office from the submitted plans of individual investigators or groups of cooperating investigators. Where a group of investigators is cooperating in managing and collecting data, a single responsible scientist should be identified for each measurement type. The Steering Committee may also review the data plan to evaluate the applicability of the data set to U.S. GLOBEC goals beyond the specific experiment.
To provide the opportunity for comparison with historical data, measurement techniques should be consistent with techniques used to collect the existing data unless there is significant scientific justification for change. When new techniques are adopted, methods for relating the new data to existing data should be developed. This requirement extends to regional comparisons as well. For example, measurements made in eastern boundary currents should be designed in consideration of the existing large database for the California Current.
Of course, remote sensing (e.g., acoustic sampling) makes it impossible to determine the physical environment at the location of the measurement. Where possible, temperature and salinity sensors should be combined with biological sensors or profiles should be taken between tows.
The reader is reminded that it is not ethical to publish data without proper attribution or coauthorship. Beyond this, the U.S. GLOBEC Scientific Steering Committee believes that the intellectual investment and time committed to the collection of a data set entitles the investigator to the fundamental benefits of the data set. Therefore, publication of descriptive or interpretive results derived immediately and directly from the data is the privilege and responsibility of the investigators who collect the data. The purpose of a data archive is to facilitate collaboration between scientists, the combination of multiple data sets for interdisciplinary and comparative studies, and the development and testing of new theories. Any scientist making substantial use of a data set should communicate with the investigators who acquired the data prior to publication and anticipate that the data collectors will be co-authors of published results. This extends to model results and to data organized for retrospective studies. As possible, the U.S. GLOBEC Data Management Office will encourage and facilitate the ethical and courteous use of data within the archive. In particular, the U.S. GLOBEC DMO will maintain a list of all data access and will notify those who access the data of our commitment to the principle that data is the intellectual property of the collecting scientists.
Data collected for U.S. GLOBEC field programs will be diverse and there is a substantial emphasis on the application of emerging technology. Therefore, the schedule for submission of data products must differentiate between types of data and provide a mechanism for flexibility where application of the data submission requirements is impractical. While these requirements must be followed, the spirit of the USGCRP Data Policy is that the data be made available whenever it is of general use. In some cases, this may require multiple submissions of the data set. This will be necessary when a portion of the data is not available promptly or if calibrations need to be changed after the original submission of the data.
Data sets consist of both the actual measurements and also descriptive data, sometimes referred to as metadata. Metadata consists of location, time, units, accuracy, precision, method of measurement or sampling, investigator, reference to publications describing the data set, a description of the processing of the data, etc. This information is often crucial for correct interpretation of the measurements. Therefore, U.S. GLOBEC databases must include all relevant metadata in a form which can be used efficiently by analyzers of the data. As the primary user of the data, the principal investigator is uniquely qualified to determine the relevant information needed to make use of the data.
U.S. GLOBEC field programs will frequently involve the coordination of several investigators making independent measurements in a cooperative sampling plan. Some information will be common to all investigators; time and location are needed for each measurement. Users of the data will need to know the full suite of measurements and the sequence in which the measurements were taken. Also, consistency of the data set is of paramount importance; measurements taken at the same time and location should have identical time and spatial coordinates recorded in the database. Of particular concern is the use of time to determine location from the navigation log. Careful maintenance of consistent timekeeping is critical and investigators are required to document the procedures which will be used to insure that temporal and spatial errors are controlled. The U.S. GLOBEC Steering Committee strongly recommends the use of a logging system which will record the underway data (navigation, and where available meteorology, near-surface temperature and salinity, and any other data collected automatically). These data should be integrated with data records made by other sampling instrumentation. This will greatly simplify the task of inventorying the data set and insuring the most accurate navigation possible. Whether or not an electronic logging system is used, responsibility for maintaining and reporting a log of all measurements lies with the chief scientist of the experiment.
6. Within three (3) months after collection, a detailed inventory of measurements made during the cruise or field season must be submitted to the U.S. GLOBEC DMO by the chief scientist of the experiment in cooperation with the participating principal investigators. This inventory will include the time and location of each measurement and a schedule for submission of full or partial data sets. Of special concern is the inventory of biological samples; all information necessary to retrieve a specific sample must be recorded in the database. Also, any anticipated problems with the data should be reported at this time.
7. Measurements which do not involve manual analysis and which would be useful to the science community must be submitted by the principal investigator within six (6) months after collection. Metadata should include any procedures that were followed to correct errors, remove noise, or otherwise modify the collected data.
Plankton samples inherently present special problems with respect to data policy. The data submission in the case of readily producible statistics, such as displacement volume, and easily producible data, such as silhouette photographs, may be available within the time frame above. The longer time frame associated with sorting plankton requires a more flexible policy which is tied to the completion of a significant portion of the sample suite from a cruise. That is, when an investigator completes the analysis of a set of samples to the degree they form a useful measure of conditions observed during a cruise, the data should be submitted. A data set becomes useful to the community at the same time that the investigator begins to use the data for ocean science. Investigators must plan resources and technician time to accomplish these primary data reduction tasks within one year from the end of the cruise during which the samples were collected.
8. All other measurements and any standard analyses of these measurements must be available to the community within one year after collection. Standard analyses include the displacement volume, species counts, and silhouette photographs of net tows, displacement volume and grain size distribution of sediment trap samples, and any other similarly producible derived data. This is not a requirement that these standard analyses be conducted. Principal investigators are responsible for selection of the types of analyses appropriate for the scientific objectives of the experiment. We expect that these analyses will be specified in the proposal and in the planning document described in policy statement 1. Any analysis similar to those listed above and produced from U.S. GLOBEC samples by the investigator or by any other scientist must be submitted to the U.S. GLOBEC DMO. Metadata must include any procedures that were followed to analyze the samples, correct errors, remove noise, or otherwise modify the collected data.
The primary responsibilities of the DMO will be to accept data from U.S. GLOBEC investigators,to verify the data has been properly transmitted, to report on the status of data submissions to the Program Manager and the Steering Committee, and most importantly to facilitate the interdisciplinary exchange of data. Also, the DMO will provide standards for the creation of the database particularly concerning the types of operations supported by data objects. These standards will be designed to conform with the data policy and to insure that the structure and appearance of the database is relatively consistent between separate contributions.
We believe that a useful database must support extension to new data types and be distributed. The wide range of data types expected within U.S. GLOBEC and the emphasis on technology application which will lead to new data types suggests that current database technology is insufficient. Object oriented methodology, which is currently emerging in programming languages and database implementations, appears to satisfy our need for multiple and easily extensible data types. Distributed databases have the advantage that the data collector is directly involved with creation of the database. If the database system is well designed, then the data collector may use the same system to access the data that is being used by the research community. This would be a significant improvement over the present situation.
To satisfy our objectives for a database which is distributed and which can handle arbitrary data types, U.S. GLOBEC is cooperating with JGOFS and the community efforts under the auspices of !' The Oceanography Society. These are evolving systems and important issues have not been fully resolved, however, the initial U.S. GLOBEC data system will be the JGOFS system. Important issues include cost and accessibility which will be assessed during the first year of the DMO. Each principal investigator and chief scientist should consider using the JGOFS system. Transferring the data to the DMO will be greatly simplified since all that is necessary with a distributed data system is the name and location of the database. The DMO will take responsibility for obtaining the data when investigators use the JGOFS type database.
Archival will be accomplished on two levels. The DMO will serve as the initial archive and for the length of U.S. GLOBEC, data will be available on-line from the DMO. In addition, the DMO will be cooperating with NODC to insure that the data is transferred to a permanent archive. NODC is committed to providing an accessible archive for all ocean data. When measurements are taken in foreign waters, the DMO will be responsible for communicating data reports to the State Department as required.
9. Investigators will either submit data to the Data Management Office or place it on-line as a U.S. GLOBEC distributed database. Standards for submission formats and development of the database will be specified by the DMO in support of the objectives of the data policy. The DMO will verify that the data is properly represented in the database and report on the status of data submission to the Program Manager and the Steering Committee at each Steering Committee Meeting.
10. The DMO will serve as an intermediate archival location and data source, will transfer data to the NODC, and will prepare the necessary documentation for data collected in foreign waters. The DMO will communicate the data policy to all producers and users of U.S. GLOBEC data. In particular, the rights of the data collectors, organizers, and producers of the data will be communicated to those who access the database.
11. Biological samples will be preserved following currently accepted practice for the particular contents. Sub-samples of a representative subset of the samples must be preserved in reagent grade alcohol for later genetic analysis. These samples will be retained for a period of 20 years and shared with the community as requested. Institutional representatives should be made aware that these samples must be stored for this extended period at a controlled temperature. Prior to disposal, the samples must be offered to the Smithsonian.
Some investigators may wish to be exempted from all or part of the data policy requirements. The only reason for exemption is a lack of general usefulness of the data collected and there may be data sets which are not of general usefulness within the time allotted. In these cases, the investigators should submit a request for exemption to the Program Manager and Steering Committee for review and a decision.
12. Requests for exemption from the data policy should be submitted to the Program Manager and the U.S. GLOBEC Steering Committee.