Representing soil temperature at different profile depths

[richardostler]

Question

How would I use CF conventions to name variables for soil temperature (in degrees celsius) measured at different soil depths. Should depth be encoded as part of the variable name or treated as a separate variable. From the standard names I think soil_temperature_ground may be the appropriate term, but I can see from other threads terms are being proposed with in_soil_layer, so I'm not sure about the best approach for units (degC not K) and depth representation.

[JonathanGregory]

Dear @richardostler

I agree that soil_temperature is the correct standard name. The depth of the measurement should be recorded as a vertical coordinate variable with a standard name of depth. If there's only one depth, it could be stored as a scalar coordinate variable rather than a coordinate variable with a size-one netCDF dimension (section 5.7). depth means "below the surface". In another open issue, it's being discussed whether we should clarify what surface means. If there could be snow on the ground, to be perfectly clear perhaps you might want depth_below_ground_level as the vertical coordinate. That would also be a new standard name, and follows existing patterns.

in_soil_layer is used in standard names which describe extensive properties of soil, rather than intensive, like temperature is. It is fine to use degC instead of K. It is recommended also to specify units_metadata="temperature: on_scale" (as explained in 3.1.2).

Best wishes

Jonathan