Re: [ferret_users] Ferret Array/variable class detection

[Ansley Manke <ansley.b.manke@xxxxxxxx>]

Hi,
The @ASN regridding transformation associates by subscript the points from the source grid onto the destination axis. This means it just puts the input data onto the output axis without looking at coordinate values.  It will always put the first element of the source variable into the first location on the destination axis, the second element into the second coordinate, etc.

You can use @NRST to regrid a string variable onto the output axis at the coordinate location closest to the coordinate in the source axis.  @XACT works for regridding string variables as well.

Ansley

On 12/9/2013 11:21 AM, Akshay Hegde wrote:

Sorry to interrupt to once again, I have come up with one more problem, with ASN transformation

$ f
     NOAA/PMEL TMAP
     FERRET v6.842 
     Linux 2.6.32-279.19.1.el6.x86_64 32-bit - 02/01/13
     10-Dec-13 00:39    

yes? use m1.nc , m2.nc
yes? sh d
     currently SET data sets:
    1> ./m1.nc
 name     title                             I         J         K         L         M         N
 TUM_6    Latitude                         1:1       ...       ...       ...       ...       ...
 TUM_9    Date - String                    1:1       ...       ...       ...       ...       ...
 
    2> ./m2.nc  (default)
 name     title                             I         J         K         L         M         N
 TUM_6    Latitude                         1:1       ...       ...       ...       ...       ...
 TUM_9    Date - String                    1:1       ...       ...       ...       ...       ...
 
yes? sh g tum_6[d=1]
    GRID GSX1
 name       axis              # pts   start                end
 PROFILE_NO X                   1 r   1                    1
 normal    Y
 normal    Z
 normal    T
 normal    E
 normal    F

yes? sh g tum_6[d=2]
    GRID GSI1
 name       axis              # pts   start                end
 PROFILE_NO1 X                  1 r   2                    2
 normal    Y
 normal    Z
 normal    T
 normal    E
 normal    F

yes? define axis/x=1:2:1 xax

! default regridding

yes? let out1 = tum_6[d=1,gx=xax]

! asn for strings

yes? let out2 = tum_9[d=1,gx=xax@asn]
yes? list out1,out2
             X: 0.5 to 2.5
 Column  1: OUT1 is TUM_6[D=m1,GX=XAX]
 Column  2: OUT2 is TUM_9[D=m1,GX=XAX@ASN]
          OUT1       OUT2
1   / 1:  5.457 "24-JUN-2006"
2   / 2:   .... ""     
    
yes? list/file=merge.nc/format=cdf/clobber/ilimits=1:2/i=1 out1,out2
 LISTing to file merge.nc

yes? let out1 = tum_6[d=2,gx=xax]
yes? let out2 = tum_9[d=2,gx=xax@asn]
yes? list out1,out2
             X: 0.5 to 2.5
 Column  1: OUT1 is TUM_6[D=m2,GX=XAX]
 Column  2: OUT2 is TUM_9[D=m2,GX=XAX@ASN]
          OUT1       OUT2
1   / 1:   .... "24-JUN-2010"
2   / 2:  5.058 ""        
 
yes? list/file=merge.nc/format=cdf/append/i=2 out1,out2
 LISTing to file merge.nc

yes? use merge.nc
yes? sh d
     currently SET data sets:
    1> ./merge.nc  (default)
 name     title                             I         J         K         L         M         N
 OUT1     TUM_6[D=1,GX=XAX]                1:2       ...       ...       ...       ...       ...
 OUT2     TUM_9[D=1,GX=XAX@ASN]            1:2       ...       ...       ...       ...       ...
 
yes? list out1,out2
             DATA SET: ./merge.nc
             X: 0.5 to 2.5
 Column  1: OUT1 is TUM_6[D=1,GX=XAX]
 Column  2: OUT2 is TUM_9[D=1,GX=XAX@ASN]
          OUT1       OUT2
1   / 1:  5.457 "24-JUN-2006"
2   / 2:  5.058 ""    

What could be the solution for this ? whether anything went wrong in using asn transformation ?

-Akshay

On Mon, Dec 9, 2013 at 11:00 PM, Akshay Hegde <akshay.k.hegde@xxxxxxxxx> wrote:

Thanks a lot mam.

this will solve my problem.

-Akshay

On Mon, Dec 9, 2013 at 10:52 PM, Ansley Manke <ansley.b.manke@xxxxxxxx> wrote:

Yes -

  `var,return=dtype`

will return FLOAT or CHAR.  Putting it into a symbol would likely give you the most flexibility to make use of it.

On 12/8/2013 8:54 PM, Akshay Hegde wrote:

Hi Developers,

Is there any function to check an array class ?

for example character classes defined by the POSIX standard

[[:alpha:]] --------> alphabetic characters

[[:alnum:]] -------> alphanumeric characters

[[:digit:]] ------> Numeric characters


[akshay@aix tmp]$ f
     NOAA/PMEL TMAP
     FERRET v6.85 
     Linux 2.6.18-371.1.2.el5 64-bit - 11/12/13
      9-Dec-13 09:58    

! Numbers

yes? let numerics = {1,2,3,4,5}

! Alpha

yes? let strings = {"1","2","3","4","5"}

yes? list numerics,strings
             X: 0.5 to 5.5
 Column  1: NUMERICS is {1,2,3,4,5}
 Column  2: STRINGS is {"1","2","3","4","5"}
       NUMERICS  STRINGS
1   / 1:   1.000 "1"
2   / 2:   2.000 "2"
3   / 3:   3.000 "3"
4   / 4:   4.000 "4"
5   / 5:   5.000 "5"

yes? list/file=tester.nc/format=cdf/clobber numerics,strings
 LISTing to file tester.nc

yes? can var/all

yes? use tester.nc

yes? sh d
     currently SET data sets:
    1> ./tester.nc  (default)
 name     title                             I         J         K         L         M         N
 NUMERICS {1,2,3,4,5}                      1:5       ...       ...       ...       ...       ...
 STRINGS  {"1","2","3","4","5"}            1:5       ...       ...       ...       ...       ...
 
yes? list numerics,strings
             DATA SET: ./tester.nc
             X: 0.5 to 5.5
 Column  1: NUMERICS is {1,2,3,4,5}
 Column  2: STRINGS is {"1","2","3","4","5"}
       NUMERICS  STRINGS
1   / 1:   1.000 "1"
2   / 2:   2.000 "2"
3   / 3:   3.000 "3"
4   / 4:   4.000 "4"
5   / 5:   5.000 "5"

yes? list ..varnames
             VARIABLE : ..VARNAMES
             FILENAME : tester.nc
             SUBSET   : 2 points (X)
 1   / 1:"NUMERICS"
 2   / 2:"STRINGS"

Here I want to create one array say let mask that should contain 0 if ..varnames is numeric class and 1 if its string class (alphanumeric)

I am interested to know how ferret checks internally whether an array is of string class or numeric class ? while displaying this error message "string/numeric data type mix-up"

yes? let myvar = ..varnames
yes? list myvar
             VARIABLE : ..VARNAMES
             FILENAME : tester.nc
             SUBSET   : 2 points (X)
 1   / 1:"NUMERICS"
 2   / 2:"STRINGS"

I need to create mask, so that if I list myvar and mask, I should get following result

Expected result :

 list myvar,myvar
             DATA SET: ./tester.nc
 Column  1: MYVAR is ..VARNAMES
 Column  2: mask is some expre
             MYVAR     mask
1   / 1: "NUMERICS" 0
2   / 2: "STRINGS"   1

 

Thanks and Regards,
Akshay Hegde
CSIR-National Institute of Oceanography,
Dona Paula, Panjim, Goa -403004