Re: About seasonal averaging

Dear Dennis,

Thank you so much for the help!!!

The dimensions of my dataset are (time, lat, lon). And the function
'month_to_season' is really useful! Thank you! I guess this function is
not listed in http://www.ncl.ucar.edu/Applications/index.shtml , under the
'climatology' examples.

And now I understand where the dimension 'month', in your example, came
from! It was from the function 'clmMonTLL'!

Once again, thank you so much!

Have a great day!

Michel

> Hello,
>
> [1]
> You say the dimensions are (lat, lon, time).
> Is that how NCL is 'seeing' the array?
>
> Do a "ncdump -h" or, equivalently "ncl_filedump"
> on the file. What order do you see?
>
> You could also do
>
> f = addfile(....)
> sst = f->SST
> printVarSummary(sst) ; What order do you see?
>
> [2]
> The climo_6 example you refer, the data is ordered (time,lat,lon)
>
> The clmMonTLL function created the named dimension "month".
> http://www.ncl.ucar.edu/Document/Functions/Contributed/clmMonTLL.shtml
> sstClm = clmMonTLL( sst) ; (month,lat,lon)
>
> The subsequent 6-month seasonal average used the "month"
> dimension name.
>
> [3]
> In your case, I would suggest "month-to-season"
>
> http://www.ncl.ucar.edu/Document/Functions/Contributed/month_to_season.shtml
>
>
> Note ... this requires full years of data .... (1948-2007) in
> your case.
>
> xDJF = month_to_season (sst(0:N-1),:,:), "DJF")
> xJJA = month_to_season (sst(0:N-1),:,:), "JJA")
>
> where N is the number of months 60*12
>
>
>
> Michel dos Santos Mesquita wrote:
>> Hello,
>>
>> I am using some SST data to produce seasonal plots (SKT from NCEP/NCAR
>> Reanalysis from 1948-2008). I would like to use the function
>> 'dim_avg_Wrap' for that. I am producing plots for the seasons DJF, MAM,
>> JJA and SON.
>>
>> There is an example from the NCL website that I am using for my plot:
>> http://www.ncl.ucar.edu/Applications/Scripts/climo_6.ncl
>>
>> But I have a question there related to the following lines:
>> *****
>> season = (/ (/ 5, 6, 7, 8, 9,10/) \ ; May-Oct [summer]
>> (/ 1, 2, 3, 4,11,12/) /) ; Nov-Apr [winter]
>> i_season = season - 1 ; NCL indices
>> etc...
>> do ns=0,1 ; 2 seasons seasonal climatology
>> sstSeaClm = dim_avg_Wrap( sstClm(lat|:,lon|:,month|i_season(ns,:)) )
>> etc...
>> *****
>> The fact that the dimension 'month' was used in the line corresponding
>> to
>> 'sstSeaClm' assumes that the dataset has a month dimension. The data set
>> I
>> am working with has a dimensions of (lat, lon, time). The time
>> dimentions
>> I am splitting into year, month, day, etc... using the ut_calendar
>> function. Do I need to assign 'month' as a dimension to my dataset to be
>> able to do seasonal averages using dim_avg_Wrap?
>>
>> In my script, I have something like this:
>> ***
>> etc...
>> season = (/1,2,12/) ; DJF
>> i_season = season-1
>> djf_clim = dim_avg_Wrap(sst(lat|:,lon|:,time|i_season(:)))
>> etc...
>> ***
>>
>> Since 'month' is not a dimension in my dataset, I used the variable
>> 'time'. How does 'dim_avg_Wrap' average my dataset in this case? Does it
>> use the indices 1, 2 and 12 from the time variable? Or does it try to
>> look
>> for the values of 1,2 and 12? Or is it necessary to add the month
>> dimension to the dataset so that the numbers 1,2 and 12 can be averaged
>> correctly? If dim_avg_Wrap uses the numbers in the variable 'season' as
>> indices, and if it averages over those indices for each year, then the
>> calculation I am doing is correct, since I have monthly data.
>>
>> I thank you in advance!
>>
>> Regards,
>>
>> Michel
>>
>

-- 
Michel Mesquita
PhD Student
UAF/IARC
michel_at_iarc.uaf.edu
-----------------------------------------
This email was sent using SquirrelMail.
   "Webmail for nuts!"
http://squirrelmail.org/
_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk