Suffix for accumulated quantities over a certain time period

[climbfuji]

In the UFS, there are many CCPP standard names for accumulated quantities over a certain period of time ("in a bucket"), for example https://github.com/NOAA-EMC/ufsatm/blob/b667c11bc8a29d83b667fd8d5a8fb32cba5426ea/ccpp/data/GFS_typedefs.meta#L9034.

  standard_name = accumulated_lwe_thickness_of_precipitation_amount_in_bucket

To understand the meaning of the variable, one then has to look at another variable, for example https://github.com/NOAA-EMC/ufsatm/blob/b667c11bc8a29d83b667fd8d5a8fb32cba5426ea/ccpp/data/GFS_typedefs.meta#L3470 or ...

standard_name = period_of_diagnostics_reset

and obtain its value and units.

Wouldn't it make more sense to specify the accumulation period explicitly in the standard name of the accumulated variable, and use a good standard notation for it, for example the ISO 8601 Duration Format?

  standard_name = accumulated_lwe_thickness_of_precipitation_amount_over_PT6H
  standard_name = accumulated_lwe_thickness_of_precipitation_amount_over_PT2H30M

[gold2718]

The issue I have with this is that the accumulation time is often a run-time value.
For instance, in the middle of a long run, we might shift to high-frequency output for some months and then shift back to daily or monthly averaging.

[climbfuji]

The issue I have with this is that the accumulation time is often a run-time value. For instance, in the middle of a long run, we might shift to high-frequency output for some months and then shift back to daily or monthly averaging.

That is unresolved by the UFS approach as well, since the bucket accumulation is usually a namelist variable (single constant). For NWP, we need to be able to output quantities at different accumulation periods (e.g. rain over 3 hours, runoff over 1 hr, ...) and have that information readily available from that variable's standard name.

[climbfuji]

I think your use case requires a different standard name suffix, but both types of suffices should be valid and documented in the standard names:

_over_FIXED_DURATION # here I am arguing for FIXED DURATION being in iso format
_over_[add some qualifier here that describes where to find the accumulation periods?]