Re: difference between avg and dim_avg (BIG DIFFERENCE ????????)

hi Sanjiv,
 the results are different probably because you have some undefined (missing or fill) values in your data set. Since you've mentioned about mask file of river basin I'm guessing so. In this case, the avg() and dim_avg(dim_avg()) would give two different results because avg() gives equal weights to all the valid values (does the averaging in a single go). But in the other method, the first dim_avg() will create a vector averaging all valid values along the mentioned dimension (in this case the last one) and then the second dim_avg() will average over those averaged values. In this way, weight given to every element is not same because after the first dim_avg() information about how many valid values were there along the first averaged dimension was lost. You can see this in the following simple example:

a = (/(/3,4,5/),(/-9,6,7/)/)
 a@_FillValue = -9
 print(dim_avg(dim_avg(a)))
 print(avg(a))

The results are 5.25 and 5, respectively. By simple calculation we get the right answer to be 5 (given by avg).

When you need to do average over an area, you can use wgt_areaave(), where you can specify weights optionally (that vary with latitude), or can use the avg function the way you used.

Hope this helps.

regards,

arindam

-- 
ARINDAM CHAKRABORTY
---------- Original Message -----------
From: Sanjiv Kumar <sanjivconscious@gmail.com> 
To: ncl-talk USERS <ncl-talk@ucar.edu> 
Sent: Tue, 2 Nov 2010 11:54:42 -0400 
Subject: difference between avg and dim_avg (BIG DIFFERENCE ????????)
> Hello,
>  
> I am using follwing functions to calculate average monthly value for the mississipi river basin.
> I have a mask file for the mississippi river basin that i used to mask the data. 
>  
> PrintVarSummary(pcp)
> Variable: pcp
> Type: float
> Total Size: 41880240 bytes
>             10470060 values
> Number of Dimensions: 3
> Dimensions and sizes:   [month | 12] x [lat | 621] x [lon | 1405]
> Coordinates: 
>             month: [0..11]
>             lat: [49.9375..24.0625]
>             lon: [-125.0208..-66.47912]
> Number Of Attributes: 5
>   units :       mm/month
>   long_name :   monthly_prism_pcp
>   _FillValue :  -9999
>   time_op_ncl : Climatology: 25 years
>   info :        function clmMonLLT: contributed.ncl
>  
> option 1:  print(dim_avg_Wrap(dim_avg_Wrap(pcp)))
>  
> (0)     58.64856
> (1)     58.87543
> (2)     74.17915
> (3)     79.49629
> (4)     102.1536
> (5)     105.3508
> (6)     89.25518
> (7)     77.08332
> (8)     74.19021
> (9)     71.37045
> (10)    71.6111
> (11)    63.43136
>  
> Average: 77.14
>  
>  
> Option2: 
>  
> do i= 0, 11
>      print(avg(pcp(month|i, lat|:, lon|:)))
> end do
>  
> (0)     40.91084
> (0)     43.56288
> (0)     61.70639
> (0)     72.314
> (0)     98.81877
> (0)     95.28362
> (0)     84.07236
> (0)     72.15498
> (0)     67.86727
> (0)     61.63789
> (0)     59.42875
> (0)     48.38778
>  
> Average: 67.16
>  
> My question is whay there is big difference between these two???????????????
>  
> Results obtained from option 2 seem to be more realistic to me.
>  
> Let me know if you need any further detail.
>  
> thanks
> sanjiv
>  
>  
> -- 
> This message has been scanned for viruses and
> dangerous content byMailScanner, and is
> believed to be clean.
------- End of Original Message -------
 
-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.

_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Tue Nov 2 11:31:35 2010