Precipitation needs to be adjusted to period

[norulz]

APCP with a 3 hour time step grib is currently shown in XyGrib as mm/h when it is in fact mm/3hr.

The suggested fix should be to divide the grib value (which is mm of precipitation accumulated in the period) by the period to give average mm/hr over the period. This will allow for one unit with one colour scale.

[dominiquehausser]

If it is easy, that's the best and comparable with usual data provided by metoffices.

[norulz]

@did-g @dominiquehausser

Some thoughts on this...

Various grib files contain at least one of three options regarding measures of precipitation:

• TOTAL precipitation for the period. This is measured in kg/m^2 (i.e the period amount usually in mm)
• RATE instantaneous rate of fall measured in mm/hr or other time period
• ACCUMULATED for the forecast increasing at every time step that has precipitation. This gives a "storm" total in mm

Currently XyGrib expresses precipitation as RATE showing mm/hr. This is only correct when the time step is one hour or else the grib file is giving RATE and not TOTAL.

I feel that XyGrib needs to cope with each of the possible inputs types above, and then to calculate the other two based on the input type. Presentation of precipitation will then be a user selected preference of Total/Rate/Accumulation. This should apply to both precipitation.

SNOW is a special case as SNOD - accumulated depth of snow is a function of not only fall, but run off and melting. So, it is not possible to calculate fall and accumulation, one from the other. No change is required here.

[did-g]

Hi

Are you sure about that? In my understanding Xygrib is already trying to convert whatever it get to mm/hr. cf.:
GribReader::computeAccumulationRecords

Have you a file wrongly decoded around ?

Regards
Didier

[norulz]

Try this one.

norm.zip

[did-g]

Ha, it's a TODO. :(
We are reaching the limit of shoehorning GRIB2 time range in grib1 logic