Describe the issue
Following a report from Laurie regarding unexpected units in several CMORised TIPMIP biogeochemical variables, we investigated the issue and identified the root cause.
The core issue is that the raw data units in the TIPMIP dataset are different from those in the ACCESS-ESM1_6 raw data that our current mappings and unit conversions were originally designed and tested against.
As a result, the CMORisation workflow applies incorrect assumptions about the source units when processing TIPMIP data, leading to mismatches between the CMORised data values and their reported units.
Variables affected
3D biogeochemical variables
The following variables are affected:
Current mapping assumptions:
| Variable |
ACCESS-ESM1_6 raw data units |
TIPMIP raw data units |
| o2 |
mol/kg |
mmol/m^3 |
| no3 |
mol/kg |
mmol/m^3 |
| dic |
mol/kg |
mmol/m^3 |
| alk |
mol/kg |
mmol/m^3 |
The current mappings and unit conversion logic were implemented assuming the ACCESS-ESM1_6 raw data convention (mol/kg), which is not valid for TIPMIP raw data (mmol/m^3).
This causes the CMORised TIPMIP outputs to contain incorrect numerical values relative to the reported CMIP units.
Productivity variable
| Variable |
ACCESS-ESM1_6 raw data units |
TIPMIP raw data units |
| pprod_gross_2d |
mol m-2 s-1 |
mmolN/m^2/s |
Similarly, the mapping for pprod_gross_2d assumes the ACCESS-ESM1_6 convention and therefore does not correctly convert the TIPMIP source data units.
Potential solution
Possible approaches include:
- Add dataset-specific unit handling logic for TIPMIP
- Introduce explicit source-unit validation before CMORisation
- Implement automatic unit conversion based on detected raw-data units
- Separate mappings/configurations for ACCESS-ESM1_6 and TIPMIP datasets
Additional context
Examples reported by Laurie:
o2 values appear to require division by 1000dic, alk, and no3 also show inconsistent magnitudespprod_gross_2d metadata does not match source data units
This issue may affect additional biogeochemical variables and requires a broader audit of TIPMIP unit conventions.
Describe the issue
Following a report from Laurie regarding unexpected units in several CMORised TIPMIP biogeochemical variables, we investigated the issue and identified the root cause.
The core issue is that the raw data units in the TIPMIP dataset are different from those in the ACCESS-ESM1_6 raw data that our current mappings and unit conversions were originally designed and tested against.
As a result, the CMORisation workflow applies incorrect assumptions about the source units when processing TIPMIP data, leading to mismatches between the CMORised data values and their reported units.
Variables affected
3D biogeochemical variables
The following variables are affected:
o2no3dicalkCurrent mapping assumptions:
The current mappings and unit conversion logic were implemented assuming the ACCESS-ESM1_6 raw data convention (
mol/kg), which is not valid for TIPMIP raw data (mmol/m^3).This causes the CMORised TIPMIP outputs to contain incorrect numerical values relative to the reported CMIP units.
Productivity variable
Similarly, the mapping for
pprod_gross_2dassumes the ACCESS-ESM1_6 convention and therefore does not correctly convert the TIPMIP source data units.Potential solution
Possible approaches include:
Additional context
Examples reported by Laurie:
o2values appear to require division by1000dic,alk, andno3also show inconsistent magnitudespprod_gross_2dmetadata does not match source data unitsThis issue may affect additional biogeochemical variables and requires a broader audit of TIPMIP unit conventions.