Re: a question about wgt_areaave_Wrap (wgt_areaave)

From: Dennis Shea <shea_at_nyahnyahspammersnyahnyah>
Date: Sun, 27 Jul 2008 21:39:10 -0600

As Saji indicates, you must compare apples-to-apples.
The NCL example you provide below computes an unweighted.

(Note the input weights are "1.0,1.0").

http://www.ncl.ucar.edu/Document/Functions/Built-in/wgt_areaave.shtml

See Example 1

Good luck

Saji N Hameed wrote:
> Hi Song Feng,
>
> First, you need to know what scheme GrADs uses - I mean what kind
> of weighted average it does ... probably, it is weighted with cosine of
> latitude???
>
> It is possible that you need to do a weighted average of some
> sort to compare with the values produced by GrADs.
>
> If you see example 1 in
> http://www.ncl.ucar.edu/Document/Functions/Built-in/wgt_areaave.shtml
> you can find ways to do various kinds of weighted averages. One of them
> should provide similar values to the GrADs produced values..
>
> saji
>
> * Song Feng <sfeng2_at_unlnotes.unl.edu> [2008-07-27 10:41:55 -0500]:
>
>> Hi All:
>>
>> I am computing the regional average using wgt_areaave_Wrap with no
>> weighting. The following are part of the code I used:
>>
>> begin
>> ;*******************************************
>> ; open file and read in data
>> ;*******************************************
>> f = addfile ("out_2.5x2.5.nc", "r") ; the data are in
>> standard 2.5 lon by 2.5 lat gridpoint.
>> t = f->pr(:,:,:) ! read the precipitation
>>
>> ; Compute the regional average over (30-50N, 90-110W). No weighting was
>> applied.
>> ; tclidiv01 = wgt_areaave_Wrap(t(:,{30:50},{250:270}),1.0,1.0, 0)
>> tclidiv01 = wgt_areaave_Wrap(t(:,48:56,100:108),1.0,1.0, 0)
>> asciiwrite ("Great.Plains.txt" , sprintf("%9.3f", tclidiv01*86400 ))
>>
>> However the regional average computed by NCL and the GrADS (using
>> ave(ave(pr,x=101,x=109),y=49,y=57)) are different. See the following for
>> an example:
>>
>> Month, Pr (output by NCL), Pr(output by GrADS).
>> 1 1.535 1.748
>> 2 1.718 1.905
>> 3 1.762 1.740
>> 4 1.873 1.732
>> 5 2.307 2.249
>> 6 2.753 2.773
>> 7 2.108 2.032
>> 8 0.961 1.094
>> 9 1.064 1.279
>> 10 1.446 1.560
>> 11 1.738 1.869
>> 12 1.465 1.685
>>
>> I used the GrADS many times, so I am pretty sure the regional average
>> output by GrADS are correct. So, what method the NCL used to compute the
>> regional average? Is he NCL compute the regional average using certain
>> kind of weighting (even I specified no weighting)?
>>
>> Best regards!
>>
>> Song Feng
>>
>
>> _______________________________________________
>> ncl-talk mailing list
>> ncl-talk_at_ucar.edu
>> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
>
_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Sun Jul 27 2008 - 21:39:10 MDT

This archive was generated by hypermail 2.2.0 : Mon Jul 28 2008 - 09:07:16 MDT