03 December 2021

A New Vision for Management of Oceanographic Data and Information

Welcome to the BCO-DMO Revolution

bcodmo_logo

Welcome to our quiet revolution. When the BCO-DMO data repository was founded in 2006, it was somewhat of an arranged marriage and a novel experiment for the time: combine two existing, well-functioning, data management offices from two different, long-standing, oceanographic research programs with a goal of gaining an economy of scale for NSF OCE’s data management needs. Turns out- it worked incredibly well. The BCO-DMO project team has been successfully collaborating with the oceanographic community ever since to help share its research output. An activity that has created a rich collection of marine and related data and information available for reuse in new scientific research.

The original system created for BCO-DMO was designed to accommodate the research data of the times- small, consistently-formatted, and locally-managed datasets, such as cruise bottle and CTD files. The data were tabular, oftentimes shared via Excel spreadsheets. The overall size of the data collection was also much more manageable back then. But, today’s rapidly evolving research techniques and instruments are producing data that challenge this original paradigm; either through high volumes, complex types, or their distributed nature across multiple locations. BCO-DMO’s underlying system needs and catalog size began to outgrow its early strategies for serving, browsing and searching the catalog. With over 1,000 projects today, and a growth in dataset holdings of more than 60% over the last five years, browsing and using keyword searches to discover data had become arduous tasks.

bcodmo_datasets_overTime

So, two years ago, BCO-DMO entered a very deliberate and rapid evolution, driven by the need to maintain the successful system and services built over the past fifteen years, but also to adapt to emerging needs of our research community by leveraging new technologies and best practices in data curation and publication.

Learn more about the FAIR Guiding Principles for scientific data management and stewardship
Learn more about the FAIR Guiding Principles for scientific data management and stewardship

Our approach to addressing these challenges is adaptive, to respond not only to where the various data types are stored (e.g., links to external raw DNA sequence or mass spectra data repositories), but also addresses their need for better documentation that will align with today’s research data stewardship demands. To ensure that BCO-DMO’s highly valuable oceanographic datasets are FAIR (Findable, Accessible, Interoperable, and Reusable, Wilkinson et al., 2016 ) for both human and machine use, we began rebuilding the entire system for flexibility in the submission, curation, and discovery of oceanographic data. This activity will increase our internal data processing efficiency and accuracy, leading to improved interoperability of BCO-DMO datasets and ultimately faster time to data publication. A key technological component of this effort is the adoption of a knowledge graph that will allow BCO-DMO to grow in a much more cost effective way. Knowledge graphs are powerful semantic tools in the quest to build a robust and flexible FAIR data system (and a great topic for another blog).

It’s a big departure from our old system, and the cliche that change is hard holds true even for repositories; but taking these hard steps now will result in a system that’s more effective, efficient, sustainable, and most importantly, easy to use. This concept of adaptive data management will underpin all our operations and forms the basis for our vision that we share below. Throughout our system redesign, we’ll continue our core operations to accept, process, and publish oceanographic research output by working closely with the individual scientists producing it; and as always, our activities continue to be framed in the context of the data lifecycle, with upcoming innovations aimed at supporting one or more of its phases. We’re excited to share some of our new system’s components and the data lifecycle stages they support; sneak peeks provided here…


Acquisition: Beginning with data acquisition, we are collaborating with the community to create common data type models that can serve as templates that can help researchers early on in their data collection to ensure their data are well organized, described, and formatted for sharing. We plan to release one in each reporting year of our current grant, and our first successful collaboration was with ocean proteomics researchers.

Our new submission tool is being user-tested now!
Our new submission tool is being user-tested now!

Contribution: To make it as easy as possible for our research community to submit their data to us, we are implementing an online submission system to streamline the process of contributing data (stay tuned for its upcoming blog entry). The system serves as an alternative option to our current historical workflow, which involves PIs completing text-based metadata forms and emailing forms and data files to us via a dedicated address. Having multiple pathways for researchers to submit their data helps to further lower barriers to sharing data. This service is self-guided and web-based, allowing data contributors to upload their data and supplemental documentation, and pre-populate new metadata or copy relevant metadata from a previous contribution to a new one. Additional planned features will allow high level quality control on tabular data. Our overall goals being to save time for submitters, help expedite data processing, and provide transparency throughout the data submission process. You can check out this new system at: submit.bco-dmo.org using your ORCiD to log in (and can easily create your ORCiD at the login page if you don’t yet have one).

Publication: Preparing data for publishing is a task that underpins our role as a repository. In order to more effectively process data and information for publication, we recently implemented a web application called Laminar that creates Frictionlessdata Data Package Pipelines to help our data managers work more efficiently and consistently, while recording the provenance of their activities to support reproducibility. You can read more about this tool that resulted from our great collaboration with the Open Knowledge Foundation in our blog post: Frictionless Data Pipelines for Ocean Science from 2020.02.09.

Discovery: An effective website is a main avenue of engagement for many organizations, but our previous site was becoming outdated and challenging to maintain. So, after much research and a fruitful partnership with Element84 , we’re scheduled to soft launch our new website soon, which will be the keystone for enabling both human and machine access to BCO-DMO data and information. This new site has been designed with data contributors, data searchers, and proposal developers in mind. It will highlight data citation as a key theme for dataset page layout, and data discovery will focus on utilizing the multi-faceted inspection capability of our new knowledge graph. Users can search and browse through data-driven facets that enable improved filtering, and enhanced keyword searches.

Stay tuned for the launch of our new BCO-DMO website!
Stay tuned for the launch of our new BCO-DMO website!

Access and Reuse: We’re keenly aware that our success as a repository depends critically on our user community’s ability to access and ultimately reuse the data we serve. So, data access in our new system will be achieved through multiple ways in order to accommodate our end-users’ needs. Data and related files will now be available right from the individual data pages via direct download. But, to provide enhanced options for data reuse (think: subsetting, filtering, and even automated ingest for scripting-savvy users) we have implemented the open-source NOAA-supported, ERDDAP data server (Simons, 2015). ERDDAP is a web application that provides features such as subsetting, filtering, data conversion to dozens of file formats, and even machine actionable URLs for automated filtering, data conversion, and viewing of data and metadata. ERDDAP does this while not imposing restrictions on how or where data are stored.

We also received input from the community that downloading multiple datasets was an important feature that we currently lack. To respond to this need, we plan to implement a dataset “shopping cart” to allow downloading of multiple datasets at once. This service will be able to calculate download sizes, informing the user of any restrictions such as size limitations before any download takes place.

Capacity Building: We believe that as curators of data, repositories can play a pivotal role in building data management capacity within the research community. So, we’re re-envisioning our outreach and education efforts, collaborating with individual scientists, projects, and organizations to support data management literacy within the oceanographic research community. We can also serve as an effective vehicle for cross-pollination of domain science and technical skills among undergraduate and graduate students and have already worked with our first undergraduate intern. We’ve started to collaborate with teachers and professors to gather needs and develop training materials, and plan to host published data analysis modules, and notebooks for use with BCO-DMO data holdings. We’re working on a communication strategy that will help socialize scientists to the resources and opportunities we provide, and we began exhibiting last year at professional conferences like Ocean Sciences, to provide additional opportunities for engagement and assist researchers in solving data challenges.


All of these activities together, feed the necessary evolution that will help BCO-DMO achieve its vision of:
An unparalleled data catalog of well-documented, interoperable oceanographic data and information, openly accessible to all end-users through an intuitive web-based interface for the purposes of advancing marine research, education, and policy.

Over the past year and a half, we’ve made great strides toward realizing this vision, and will soon be implementing the components of our new system; many will be transparent to our users, but others will be a significant departure (for the better!) from the system that our community has helped build with us. In the coming months both data submitters and end users will see a vastly updated system that improves the acts of data sharing, discovery, and access. To help introduce the components of this new system, we plan to announce changes through Twitter and blog posts like this one; not only to inform the community of pending changes, but to provide opportunities for feedback. So be sure to follow us on Twitter , check out our blog , and even look for our BCO-DMO related posts in the OCB newsletter !


Last modified: 2022-03-17 15:30:00

Comments

Have questions?

Have questions?

CONTACT US
TOP