Re: [ferret_users] different add_offset/scale_factor for different step files

[Ansley Manke <a.c.manke@xxxxxxxxx>]

Hi Seon,

This is correctly handled by PyFerret. The the add_offset and scale_factors from each step file are applied as the data are read from the file. This is noted in the documentation about descriptor files

I have attached a tar file with some simple example files, vary_scale.des which uses stepfiles varyscale_1.nc and varyscale_2.nc.  Here is what happens:

> ! First open each stepfile.  The
>         variable tt has the same data values in both, but different
>         scale and offset attributes:
>         
>         yes? use varyscale_1.nc
>       yes? show attributes tt
>            attributes for dataset:
>         ./varyscale_1.nc
>        TT.missing_value = -1.E+34
>        TT._FillValue = -1.E+34
>        TT.long_name = 1 + 0*T[GT=DAYT] 
>        TT.scale_factor = -1
>        TT.add_offset = 0.5
>       yes? list tt
>                    VARIABLE : 1 + 0*T[GT=DAYT]
>                    FILENAME : varyscale_1.nc
>                    SUBSET   : 10 points (TIME)
>        01-JAN-1990 00 /  1: -0.500
>        02-JAN-1990 00 /  2: -1.500
>        03-JAN-1990 00 /  3: -2.500
>        04-JAN-1990 00 /  4: -3.500
>        05-JAN-1990 00 /  5: -4.500
>        06-JAN-1990 00 /  6: -5.500
>        07-JAN-1990 00 /  7: -6.500
>        08-JAN-1990 00 /  8: -7.500
>        09-JAN-1990 00 /  9: -8.500
>        10-JAN-1990 00 / 10: -9.500
>       
>       yes? use varyscale_2.nc 
>       yes? show attributes tt
>            attributes for dataset:
>         ./varyscale_2.nc
>        TT.missing_value = -1.E+34
>        TT._FillValue = -1.E+34
>        TT.long_name = 1 + 0*T[GT=DAYT] 
>        TT.scale_factor = 2
>        TT.add_offset = -0.02
>       yes? list tt
>                    VARIABLE : 1 + 0*T[GT=DAYT]
>                    FILENAME : varyscale_2.nc
>                    SUBSET   : 10 points (TIME)
>        11-JAN-1990 00 /  1:   1.98
>        12-JAN-1990 00 /  2:   3.98
>        13-JAN-1990 00 /  3:   5.98
>        14-JAN-1990 00 /  4:   7.98
>        15-JAN-1990 00 /  5:   9.98
>        16-JAN-1990 00 /  6:  11.98
>        17-JAN-1990 00 /  7:  13.98
>        18-JAN-1990 00 /  8:  15.98
>        19-JAN-1990 00 /  9:  17.98
>        20-JAN-1990 00 / 10:  19.98
>       
>       ! Now open the descriptor file and read the
>         data. The scaling is
>         ! applied for each step file.
>       
>       yes? cancel data/all
>         
>       yes? use vary_scale.des
>       yes? list tt
>                    VARIABLE : 1 + 0*T[GT=DAYT]
>                    DATA SET : Model Output, Daily
>         Averages
>                    FILENAME : vary_scale.des
>                    SUBSET   : 20 points (TIME)
>        01-JAN-1990 00 /  1:  -0.50
>        02-JAN-1990 00 /  2:  -1.50
>        03-JAN-1990 00 /  3:  -2.50
>        04-JAN-1990 00 /  4:  -3.50
>        05-JAN-1990 00 /  5:  -4.50
>        06-JAN-1990 00 /  6:  -5.50
>        07-JAN-1990 00 /  7:  -6.50
>        08-JAN-1990 00 /  8:  -7.50
>        09-JAN-1990 00 /  9:  -8.50
>        10-JAN-1990 00 / 10:  -9.50
>        11-JAN-1990 00 / 11:   1.98
>        12-JAN-1990 00 / 12:   3.98
>        13-JAN-1990 00 / 13:   5.98
>        14-JAN-1990 00 / 14:   7.98
>        15-JAN-1990 00 / 15:   9.98
>        16-JAN-1990 00 / 16:  11.98
>        17-JAN-1990 00 / 17:  13.98
>        18-JAN-1990 00 / 18:  15.98
>        19-JAN-1990 00 / 19:  17.98
>        20-JAN-1990 00 / 20:  19.98
>     
    
    When looking at the descriptor file, If you ask for the values of
    the add_offset and scale_factor attributes, the values for the first
    file are shown, but when loading data, the attribute values from
    each file are applied.

On 8/25/2022 4:36 AM, Seon Tae Kim wrote:

Dear ferreters 

When I tried to use a descriptor file to aggregate the multiple netcdf files, 

I realized that the add_offset and scale_factors in the first step file are to all step files in a multi-file set of a descriptor file.

I can not find any solution to this problem in the ferret e-mail archives. 

Best, 

Seon

Attachment: vary_scale_des.tar
Description: Unix tar archive