The Data Portal and the neonUtilities R package have been updated to default to downloading only data from the most recent Data Release. Provisional data, which are subject to change without notice, are still accessible, but are downloadable on an opt-in basis, rather than opt-out.
To balance the needs for timely data publication and reproducibility, NEON initially publishes new data as Provisional, and annually generates a formal Data Release for each data product. The Provisional data may be updated or reprocessed at any time. Data Releases will not change and each Product Release is tagged with a DOI. For further details about the Provisional and Release data process, see the web page for Data Product Revisions and Releases.
Wherever possible, NEON encourages users to run their analyses with Released data. Data product Releases can be cited with their DOIs and are unchanging over time, greatly improving reproducibility. However, some analyses rely on access to the most recent data possible (such as forecasting), and in those cases the increased uncertainty associated with Provisional data is an acceptable trade-off. If you must use Provisional data, we recommend storing a version in a repository to accompany any scientific products generated with these data.
This is a breaking change for some neonUtilities downloads. Download function calls that previously returned Provisional data will return only Released data unless the function call is modified. In the neonUtilities R package download functions, the new `include.provisional` input parameter defaults to FALSE. Add include.provisional=TRUE to the function call to access Provisional data.
On the Data Portal download interface, a new selection step has been added, with a checkbox to indicate whether to include Provisional data files in the download. The default setting is not to include Provisional data.
This change went live on the Data Portal and in the CRAN release of neonUtilities (version 2.4.0) on October 20, 2023.