C3S seasonal GRIB2 encoding issues

ecCodes doesn't allow writing some parameters in GRIB2 (e.g. param=49) regardless a definition exists for them in the param DB

(possible cause? 49 not present in paraId.def file)

Sea-ice temperature (param=260649) can't be used: restricted to some origins?

ECCODES ERROR   :  concept: no match for paramId=260649

ECCODES ERROR   :  concept: input handle edition=2, centre=ecmf

ECCODES ERROR   :  Please check the Parameter Database 'https://apps.ecmwf.int/codes/grib/param-db/?id=260649'

It is currently in use for CARRA/CERRA 

Is this the right paramId to use? what about the recently added 262010?

Snow-depth:

param=141 (used by C3S in GRIB1) produces "unknown" name

param=228141 (used by CARRA) has different units (kg/m2 instead of m) involving a factor of 1000 (i.e. users would see different numbers depending on the GRIB edition)

Also... what typeOfFirstFixedSurface value? 1 (ground) or 103 (above ground) at 0m?

shouldn't this be included in GRIB2 parameter definition?

snow density

param=33 produces "unknown" name

Also... what typeOfFirstFixedSurface value? 1 (ground) or 103 (above ground) at 0m?

shouldn't this be included in GRIB2 parameter definition?

Radiation fluxes
(params = 147/146/175/177/179/169/212/178/179)

units in GRIB1 = J/m2; in GRIB2 = W/m2

what's the approach for accumulations in GRIB2? (from the beginning of forecast?)

Wind stresses (param = 180/181)

produces "unknown" name

(potentially same units/accumulation issue as above)

Precipitation (228/142/143)

shouldn't typeOfFirstFixedSurface=1 value be forced by the parameter definition?

• 143 (convective precipitation): name changes from GRIB1 (cp) to GRIB2 (acpcp) (the GRIB2 short name is different from the Parameters DB) 
• 142 (stratiform precipitation): yet to be fully tested/checked
• 228 (total precipitation) produces "unknown" name

snowfall (param 144) and evaporation (param 182)

produce "unknown" name

runoff (8/9/205)

• surface runoff (8) produces "unknown" name
• sub-surface runoff (9) produces "unknowon" Discipline, etc...
  • what typeOfFirstFixedSurface should be used here?
• total runoff (205): yet to be fully tested/checked

Hindcast encoding

• productDefinitionTemplateNumber = 60 (or 61 for interval-based)
  • include value of modelVersionDate? definition and feasibility? relation to system?
• typeOfGeneratingProcess = 4 (Ensemble) or 15 (Hindcast)?