Every second Friday 11:00am on zoom, commencing 21/02/2025.
Help please!
The program will be all the richer with community input, and maybe a starting point for new collaborations… Perhaps some of you have an interest in demonstrating or showing off one of your favorite tools, models or analysis?
Examples might include: WW3, panantarctic configs, MOM6 idealised modelling, Oceananigans, regional modelling, evaluation to observations, novel OM2 diagnostics, running isomip+ mom6 etc… The only criteria is that you think it’s of value to the community.
This program will be much richer with community input! We are especially interested to hear from community members that would be able to help facilitate a session, if this is the case, please put your “name; email address” in your suggestion, e.g., "Running and understanding key diagnostics in WOMBAT-lite (Pearse Buchanan; Pearse.Buchanan@csiro.au). Similarly, help is also appreciated for existing proposed sessions, please contact Chris Bull (chris.bull@anu.edu.au) if you’d like to help out.
At the next COSIMA meeting (12/12) we’ll poll on the suggested sessions to finalise the program. Please come along on Thursday if you’d like to participate in the program, available sessions will be based on interest.
Hi there! I was hoping to make a suggestion for the training program: metadata, what and how to record it, and any established conventions/guidelines that we should be following. I find that metadata records are not consistent across models, so I’m still trying to figure out what is good practice.
The reason for my suggestion is because I am producing a large number of files after running a marine ecosystem model and ideally I would like to include information that make it easier of potential users to understand these outputs.
Thanks for your interest @hrsdawson, we’ve recorded some of them when it has been requested and the presenter doesn’t mind. The recording system doesn’t currently work with breakout rooms so it also depends on the format of the training.
Given your interest, I’ll ask the presenter if they are open to it being recorded.
Thanks very much for your suggested topic! That is indeed a tricky space to navigate and I don’t think anyone in the ocean team would be able to present on as such. Relatedly, I’m not sure this training program is the right place for this kind of training… Of course, anyone from the community is welcome to present what they think will be of interest but from what you’ve written it sounds like it might appeal to a broader audience?
Strongly agree that a training session on metadata is a good idea. I’ve caused myself far too many issues recording inadequate metadata over the years.
From my personal perspective, the real difficulty with metadata is often poor tooling for generating it automatically - when adding metadata becomes a chore, it’s often the first thing to slip in the interests of saving time. Perhaps we could direct some effort towards building tooling to scaffold metadata - I suspect the ROI would be substantial.
For the use case @lidefi87 mentions, something like an autogenerated intake catalog wouldn’t hurt as a start - indexing is always a good place to begin with creating metadata. I don’t know whether these marine ecosystem models would require substantially different building processes to what we currently have - is a marine ecosystem model another name for an ocean biogeochemistry model or is it a completely different kettle of fish (pun intended)?
The marine ecosystem models (MEMs) I work with are different to BGC models. The aim of MEMs is to estimate changes in biomass of marine organisms and often use outputs from ocean and BGC models as forcings.
Outputs from other MEMs do not include a great deal of metadata and wanted to see what other disciplines did. Since ocean/BGC models appear to be better documented, I was hoping to get more advice from this community so I could adapt it to my needs.
I think I’m getting a better idea of what could be useful to include in the metadata after going through documentation about best practices. But I still think it would be a great topic to cover since anyone wanting to share data would likely benefit from this.