Defining an argopy policy for access to non-official Argo dataset contributions

[gmaze]

Rational

With the development of non-official Argo dataset developed by some research groups, we foresee an increase in the number of dataset made available by Argopy in the future (eg: #623 ).

We strongly welcome these contributions and believe they will help science in providing a centralised access to Argo data through Argopy (eg: intercomparison, reproducibility, simplified access, etc...).

However, the Argopy team also reached the conclusion that for clarity with regard to the end-user, non-official Argo dataset should not be made available on the same ground, ie with the same APIs, as the official Argo dataset from the GDAC servers.

The Argopy team needs to provide contributors with a clear framework to provide access to non-official Argo dataset through the Argopy library:

• a definition of products under the scope of Argopy (and what are the products that shall never be made accessible with Argopy),
• a definition of APIs methods to use to provide access to non-official Argo data (method name and description of arguments).

Proposition

With ArgoFloat, the current internal status is that non-official dataset could be made available with a new method:

from argopy import ArgoFloat

ds = ArgoFloat(WMO).open_product('product-doi')

Here the open_product makes it clear compared to the regular open_dataset that user is accessing a third-party product, not official Argo GDAC data.