Re: missing_value problem about GODAS monthly data

I see the issue. This is a good example of the "golden rule of
data processing" .... Look at your data before processing.

Consider:

     b = addfile(diri + f2, "r")
     us = b->ucur(:,{5},:,:) ; us is type short
     printVarSummary(us)

Variable: us
Type: short <=============================
Total Size: 3611520 bytes
             1805760 values
Number of Dimensions: 3
Dimensions and sizes: [time | 12] x [lat | 418] x [lon | 360]
Coordinates:
             time: [35429..35763]
             lat: [-74..64.999]
             lon: [ 1..360]
Number Of Attributes: 20
   level : 5
   missing_value : 32767 <=============================
[snip]
   _FillValue : 32766 <=============================
[snip]
   add_offset : 0
   scale_factor : 0.000122074
   valid_range : ( -2, 2 )
   actual_range : ( -1.51604, 1.53069 )

---
Note the the missing_value and _FillValue attribute are different.
If you do a
%> ncdump -v ucur ucur.1998.nc | less
you will see many 32767 [land] and no 32766.
NCL only recognizes (_FillValue) as recommended by Unidata
and the Climate and Forecast Convention.
In this case, it is the user responsibility to 'manually'
make the change. For this dataset
     us@_FillValue    = us@missing_value           ; Force over write
     u    = short2flt(us)                          ; convert to float
     delete(us)                                    ; no longer needed
The attached script shows the complete "fix"
Good luck
On 04/27/2010 01:10 AM, moon wrote:
> Dear NCL users,
>
> I download GODAS monthly data from the website
> "http://www.esrl.noaa.gov/psd/data/gridded/data.godas.html".
> Draw the sea surface current over the dzdt, but the value over the land
> can't be masked.
> I think it's the problem of the missing value, delete the
> "missing_value", but it still doesn't work.
> Do you have any solution?
>
> Thank you very much.
>
> Regards
> moon
[SNIP]

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk