Update your Public Dataset

ESS-DIVE allows updates to published datasets while retaining the same DOI.

In fact, we encourage small updates after publication that enhance the data quality, such as completing related citations for recently published manuscripts, adding related identifiers from subsequent research, or expanding the keywords to improve discoverability. These updates can be made freely without further action.

For updates that modify the content of your data publication beyond this, you may be interested in versioning your publication. As a best practice, versioned publications should clearly describe the data provenance. We recommend reviewing these guidelines and expectations for versioning datasets before you update your publication.

It's important to consider the impact your change will have on the research products that have previously used and cited your data.

Options for Versioning Public Datasets

There are two options for providing updates after your dataset has been published:

  1. Update an existing dataset,

  2. Publish a new dataset with a new DOI.

These both result in a new version of the data. Consider option 1 if you are extending data, adding procedural metadata, or providing non-breaking corrections or usability enhancements to the data. Consider option 2 if you want to retain the original dataset citation or obsolete a prior data publication.

There are many other reasons why you may choose one option over the other. Ultimately, it is your decision whether the nature of the change merits a new scientific publication or not. Continue to the linked sections for recommendations on how to version datasets that clearly describe data provenance.

How will versioning affect my citation?

Changes made to metadata that appear in the citation (authorship, title, or publishing project) will change the citation. It may be your preference to create a new dataset instead of altering the citation. Or you may find that retaining the same DOI, despite any citation alterations, is valuable for the provenance of your data.

Regardless of your choice, the DOI in your citation will always remain the same.

What about updating Large Data on Tier 2?

Researchers who have published very large datasets on Tier 2 should more carefully consider whether the file volume of the change merits a new publication. Please reach out to ESS-DIVE to discuss the nature of your update.

Data Publications cannot be deleted

Formally deleting or removing public datasets is not allowed.

As a long-term data repository, ESS-DIVE cannot delete or remove data once it has been published. Data can be obsoleted by a new version as necessary, however we are not able to limit public access to open data, even if it is obsolete.

Obsoleting or retiring a public dataset on ESS-DIVE requires modifying the old dataset metadata properly and publishing the latest version as a new dataset (option 2).

1. Update an Existing Dataset

When updating an existing DOI, we recommend updating your metadata to clearly describe all changes to the publication:

  • Abstract: Add a brief statement clearly describing all changes, which files were changed/added, and the date they were made.

  • Methods: Add a methods step that describes how the dataset was updated. You may go into detail here. Include the date/year that the dataset was originally published first, followed by subsequent versions.

  • Publication Date: Update to the date or year that you changed the dataset.

  • Change log: Consider adding a simple changelog file in your dataset recording all specific changes with timestamps.

  • (If applicable) Temporal Coverage: If adding new dates, update date range to include updated data.

  • (If applicable) Authors: If applicable, add any new contributors to the author list.

  • (If applicable) Title: Consider updating any time frames or specific sites mentioned in the title which have been expanded on.

  • (Optional) Version numbers: You may prefer to include specific version numbers in your Title (e.g. “Original dataset title (Version 2.0)”) and/or File Names (e.g. "filename_v2.csv") to quickly identify newer versions.

Option 1 Examples

The most common type of updates you see in these examples are periodic updates that extend timeseries data (e.g., sensor streams or new field seasons). This is particularly useful for long-term or key project data products.

doi:10.15485/1866836; doi:10.15485/1603775; doi:10.15485/1660962

  • New data: additional dates or sampling locations; including analysis results as they become available.

  • Improved data quality: reformatting files with reporting formats; improving data and code usability; providing additional documentation or READMEs; expanding methodology.

  • Corrections: units; dates; coordinates in data files; typos.

2. Publish a New Dataset and DOI

When creating a new dataset and DOI version, we recommend the following:

  • Old Abstract: Add a statement at the beginning of the abstract in the original dataset to indicate that "A new version of this dataset is available at https://doi.org....."

  • Old File Names: Update any obsoleted dataset file names in the original dataset to indicate that they should not be used. Do not delete the files.

  • Old Related Reference: Add a full citation to the new dataset as a related reference in the original dataset.

  • New Related Reference: Add a full citation to the old dataset as a related reference in the new dataset.

Option 2 Examples

The reasons for publishing data under a new DOI are varied and can come down to preference. In general, you should get a new DOI if your changes will alter the publication substantially enough that it impairs the reproducibility of research that has previously used and cited your data.

⭐ Do you have feedback on how you would prefer versioning to work? Let us know! Email [email protected] or use our contact form.

PreviousTroubleshootingNextRegister Dataset Citations

Last updated