Depending on which ERA Interim files you have, it will be much more complicated than that.

If you have the high temporal resolution files (i.e. not the daily or monthly averages) you must be careful. If I remember correctly, there is 1 file every 12h for the precipitation fields. That file has the first guess (0h) and also four forecast fields (3h, 6h, 9h, and 12h). Those 5 time-steps, in a single 12h interval file, contain the accumulated precipitation at that time, not the instantaneous nor the 3h average precipitation!

Therefore, if you want to know how much precipitation there was in a day you'll have to add the 12h FCT fields from the two files you have for that day! For example:

• file for 0z1jan89, fct 3h = precipitation accumulated between 0z1jan89 and 3z1jan89
• ....
• file for 0z1jan89, fct 12h = precipitation accum between 0z1jan89 and 12z1jan89
• file for 12z1jan89, fact12h = precipitation accum between 12z1jan89 and 0z2jan89
• ...

You can easily realize that ERA Interim precipitation is stored in this strange way by plotting the time series (e.g. for 15 time-steps) from a single grid point. No matter which grid-point you chose, the precipitation only increases from 0z to 12z, then again for 12z to 24z. 

Because this is an accumulated field, its units are in equivalent height (in meters) of the water layer in a 1 m2. Hence, if you take the value at 12z and want to go from accumulated meters to mm/day, you have to multiply by 1000 and divide by 12 (because it was accum between 0z and 12).

Moreover, if you want the 3h averages precipitation rates in mm/h, for instance, you have to calculate the difference between consecutive time-steps in the file. In grads it would be:

set t 2 last

precip_rate = (tp - tp(t-1)) *1000/ 3

Cheers

Henrique

Are you shure about the units, i believe ERA INTERIM was in mm, maybe you need to contact with the data provider to be shure.

Yes, I saw in the table 128 and the unit is "m".

I suppose the data is m/day then to covert to mm/day we have to multiply by 1000 (m->mm).

Hi Guilherme,

correct.

if you are using daily data, 

to covert to mm/day you are supposed to multiply by 1000.

Dear Enrico, 

Thank you for your answer.

Depending on which ERA Interim files you have, it will be much more complicated than that.

If you have the high temporal resolution files (i.e. not the daily or monthly averages) you must be careful. If I remember correctly, there is 1 file every 12h for the precipitation fields. That file has the first guess (0h) and also four forecast fields (3h, 6h, 9h, and 12h). Those 5 time-steps, in a single 12h interval file, contain the accumulated precipitation at that time, not the instantaneous nor the 3h average precipitation!

Therefore, if you want to know how much precipitation there was in a day you'll have to add the 12h FCT fields from the two files you have for that day! For example:

• file for 0z1jan89, fct 3h = precipitation accumulated between 0z1jan89 and 3z1jan89
• ....
• file for 0z1jan89, fct 12h = precipitation accum between 0z1jan89 and 12z1jan89
• file for 12z1jan89, fact12h = precipitation accum between 12z1jan89 and 0z2jan89
• ...

You can easily realize that ERA Interim precipitation is stored in this strange way by plotting the time series (e.g. for 15 time-steps) from a single grid point. No matter which grid-point you chose, the precipitation only increases from 0z to 12z, then again for 12z to 24z. 

Because this is an accumulated field, its units are in equivalent height (in meters) of the water layer in a 1 m2. Hence, if you take the value at 12z and want to go from accumulated meters to mm/day, you have to multiply by 1000 and divide by 12 (because it was accum between 0z and 12).

Moreover, if you want the 3h averages precipitation rates in mm/h, for instance, you have to calculate the difference between consecutive time-steps in the file. In grads it would be:

set t 2 last

precip_rate = (tp - tp(t-1)) *1000/ 3

Cheers

Henrique

Dear Henrique, 

Thank you for all the explanations, now things make sense.

Cheers,

Guilherme.

Dears

Could I have your help to open precipitation data of ERA-Interim via GrADS?

I've attached its "grib" and "ctl" files. However I encountered with "Cannot contour grid - all undefined values". I'm not sure about the "ctl" file and I'll be thankful if I could have your help.

Sincerely

Ehsan

Dear,

If you use CDO (some examples how to use: https://sites.google.com/site/jgmsantos/tutoriais/cdo), try:

cdo -s -f nc copy -chname,var228,tp Prec_Jan_00_12.grib Prec_Jan_00_12.nc

copy = convert grib to netcdf format

chanme = change the name var288 to tp

To convert to NetCDF format. If you open your ctl file you'll see is missing the "Prec_Jan_00_12.idx". You need this file to open your data.