Support Multiple Unit Representations During Dataset Compilation

[JessyBarrette]

Problem Statement

When ERDDAP compiles source files into a dataset, it may encounter the same variable with different unit representations across files (e.g., salinity as PSU, 1, 1e-3, 1E-03). Currently, ERDDAP would reject the files where the variable units are different, causing aggregation issues even though these representations could be considered similar. This is a common issue when aggregating multiple historical data sources within a single dataset.

Proposed Solution

Add a new element in the object within datasets.xml:

xml<dataVariable>
    <sourceName>salinity</sourceName>
    <destinationName>salinity</destinationName>
    <addAttributes>
        <att name="units">PSU</att>
    </addAttributes>
    <unitAliases>PSU,1,1e-3,1E-03</unitAliases>
</dataVariable>

During compilation, ERDDAP would accept any unit string listed in and present the unit attribute within the ERDDAP dataset. This feature would not modify the data, only approuve certain unit expresions.

Does similar functionality already exist?

I can see in the docs that messages.xml has a way to map units to another expression ... for the whole erddap server. But maybe it would be better to be variable specific.