I am running an idealized experiment with a atmospheric model with output in the format of netcdf. But the netcdf file header does not describe the variable and and dimension clearly. for example :
netcdf qc3d {
dimensions:
Time = UNLIMITED ; // (1453 currently)
bottom_top = 45 ;
south_north = 32 ;
west_east = 12288 ;
variables:
float Time(Time) ;
Time:axis = "Time" ;
Time:long_name = "Time" ;
Time:standard_name = "Time" ;
Time:units = "minutes since 1900-01-01 " ;
double zc(bottom_top) ;
zc:axis = "Z" ;
zc:long_name = "vertical height of model layers, MSL" ;
zc:standard_name = "altitude" ;
zc:units = "m" ;
zc:positive = "up" ;
double yc(south_north) ;
yc:axis = "Y" ;
yc:long_name = "y-coordinate of grid cell centers in Cartesian system" ;
yc:units = "m" ;
double xc(west_east) ;
xc:axis = "X" ;
xc:long_name = "x-coordinate of grid cell centers in Cartesian system" ;
xc:units = "m" ;
float qc(Time, bottom_top, south_north, west_east) ;
qc:long_name = "cloud water mixing ratio" ;
qc:standard_name = "cloud_water_mixing" ;
qc:units = "kg kg-1" ;
qc:_FillValue = 9.96921e+36f ;
qc:missing_value = 9.96921e+36f ;
// global attributes:
:model_tag = "CSU VVM" ;
:references = "http://kiwi.atmos.colostate.edu/pubs/joon-hee-tech_report.pdf" ;
:contact = "jung@atmos.colostate.edu" ;
:institution = "Colorado State University" ;
:VVM_casename = "GATE_PHASE_III " ;
}
there are 4 dimensions, Time(Time), zc(bottom_top), yc(south_north), xc(west_east) and 5 variable in which first 4 variables should be the dimension of the fifth variable, so called qc. But it seems not. The dimension is just a series number index from 1 to 45, 32 ...whatever else.
I would like to calculate some variable which is the function of pressure. in the CDO the script is like this
cdo expr,"pottemp=temp*((100000/clev(temp))^0.287);" ifile pottemp.nc
(this code is from here)
but I just obtain a series of index like 1 ,2 ,3 ... not the real pressure level when I use this function, clev(qv).
So how can I rewrite the variable dimension like qc from
qc(Time, bottom_top, south_north, west_east) ;
to
qc(Time, zc, yc, xc) ;
I think it's possible for me to finish this in matlab, but I just don't want to open this dataset because it's size is too large... so I am trying to find some tools like ncks or cdo, but still failed to do this.
thanks a lot.
With NCO try
ncap2 -s 'pottemp=temp*((100000/zc)^0.287)' in.nc out.nc
extending answer to cover additional question below:
first use ncap2 to explicitly set zc to the values in the ascii file:
ncap2 -s 'zc[$zc]={1.5,900,2500,3500}' in.nc out.nc
(assuming zc is size 4 dimension). then define pottemp based on that zc.