There’s a new Eos article that was just published this week about AGU’s Data Policy. It can be found here:
Written by Brooks Hanson, the Director of Publications at AGU HQ, and Rob van der Hilst, the chair of the AGU Publications Committee, it gives historical context and broader perspective on why AGU is now enforcing this policy with every manuscript that is submitted to any of its journals. I highly encourage everyone to read it.
To very quickly summarize, the reason behind the policy is reproducibility of the results and veracity of the scientific findings. I agree with the Data Policy and think that it is a worthwhile goal for which all of us should strive.
In addition, the article serves as an open call for research community members to participate in an upcoming conference on data availability and archiving. This meeting is in early October, just a few weeks away, so if you have interest in being part of this discussion, then please contact Brooks or Rob immediately.
Observational data in the field of space physics is usually available at the Coordinated Data Analysis Website, Planetary Data System, one of the World Data Centers, or a mission-specific or an instrument-specific website. These are all legitimate locations to which you can direct readers to find the data used in your study. If it is from a smaller experiment, such as a rocket or balloon flight, a short-lived field campaign, or laboratory data, then the observations used in the manuscript should be made available by the authors at a personal website, an institutional server, or even as an electronic supplement to the paper.
Numerical data is very similar to the latter case above: unless it is a CCMC simulation and the results are available through their run output repository, then the authors need to make available any code results that were used for any plot or value in the paper. Again, this can be made available at a personal website, an institutional server, or as electronic supplements to the manuscript.
The Eos article includes a couple of paragraphs about the Data Policy requirement of the availability of computer codes. If commercial software was used, then simply stating the names of those packages is sufficient. If code was written to process data or solve equations, though, the ideal would be to have all of this be open source code so others can error check the code and ensure the correctness of the calculation. Making code available is the goal, but AGU is also giving each journal editorial board discretion to implement this policy specifically for their field. The editors of JGR Space Physics would like to have all codes be open source and available, but we realize the code is your intellectual property and so we are not requiring this availability as a stipulation of publication. The output from the model, however, is “numerical data” and needs to be available for others to download (from somewhere) and examine. When you submit a paper to JGR Space Physics, you will be asked to provide a statement about the availability of modeled data as well as instrumental data.
For more on this topic, the Data Policy can be found here and I have written a couple of other posts about the policy, when it first became enforced and later to help clarify the model availability issue.