CDO

Greetings. I am new in the field of climate modelling. My region of interest is Botswana ( a Country in Southern Africa) and I am trying different dataset to see which ones can represent our past climate well before i could move into forecasting (near Future and far future) using RegCM4 my simulation is from 1981 to 2010 using EIN15 dataset. The model ran successfully for a period of 30 years and I extracted precipitation from the model output ( I merged all the _SRF..nc model output). I converted the data to lonlat since it was curvilinear using “cdo remapbil, mygrid.txt” which contained the following:

gridtype=lonlat

xsize=62

ysize=31

xfirst=8.155898

xinc=0.4866919

yfirst=-29.06826

yinc=0.4866919

I also converted the observational precipitation from “cru” dataset to the above dimension and resolution. From there I did a bit of data processing using pipping in cdo: "the idea was to sum the precipitation into monthly and also sum all the Tsize to have only 1 timestep"

cdo timmean -monsum -selmon,10,11,12 -selyear,1981/2010 Bmerge_Remapped_89_10_preip_cdo.nc pr_89_10_final_OND.nc

I did the same for the cru dataset now moving to final step to do subtraction (model output from “cru observational dataset for precipitation”) using the following command: cdo sub pr_89_10_final_JFM.nc cru_81_10_JFM_botsremapped.nc PR_FINAL_89_10_FINAL.nc

I got the following error message: cdo sub: Filling up stream1 >pr_89_10_final_JFM.nc< by copying the first variable of each timestep.

cdf_put_vara_double: name=stn type=NC_INT minval=-8999999873090293122515668784119808.000000 maxval=-2.000000

Error (cdf_put_vara_double): NetCDF: Numeric conversion not representable

Kindly assist or advice on way forward