Re: Listing weights used for "@SBX" transform...

[Ansley Manke <Ansley.B.Manke@xxxxxxxx>]

Hi Gary,
The SBX is a boxcar smoother, so it weights all the points equally.
sum from L-6 to L+6 (sum of values)/12.  Since you have an even-
length window,the weights on each end will be 0.5.
Here's a way to see what  the weights are:  Make a variable
on your
time axis which is 1 at one point, and zero everywhere else. 
Apply the
transformation to that variable.

yes? use monthly_navy_winds
yes? let temp = uwnd[x=180,y=0]
yes? let tt = if l eq 20 then 1+0*temp else 0*temp
yes? list/l=10:30 tt, tt[l=@sbx:12], tt[l=@sbx:12]* 12

             DATA SET: /home/ja9/tmap/fer_dsets/data/monthly_navy_winds.cdf
             TIME: 02-OCT-1982 13:15 to 02-JUL-1984 17:45
             LONGITUDE: 180E
             LATITUDE: 0
 Column  1: TT is IF L EQ 20 THEN 1+0*TEMP ELSE 0*TEMP
 Column  2: TT[T=@SBX:12] is IF L EQ 20 THEN 1+0*TEMP ELSE 0*TEMP (box smoothed by 12 pts on T)
 Column  3: EX#3 is TT[L=@SBX:12]* 12
                     TT       TT   EX#3
17-OCT-1982 / 10:  0.000  0.00000  0.000
17-NOV-1982 / 11:  0.000  0.00000  0.000
17-DEC-1982 / 12:  0.000  0.00000  0.000
17-JAN-1983 / 13:  0.000  0.00000  0.000
16-FEB-1983 / 14:  0.000  0.04167  0.500
18-MAR-1983 / 15:  0.000  0.08333  1.000
18-APR-1983 / 16:  0.000  0.08333  1.000
18-MAY-1983 / 17:  0.000  0.08333  1.000
18-JUN-1983 / 18:  0.000  0.08333  1.000
18-JUL-1983 / 19:  0.000  0.08333  1.000
18-AUG-1983 / 20:  1.000  0.08333  1.000
17-SEP-1983 / 21:  0.000  0.08333  1.000
18-OCT-1983 / 22:  0.000  0.08333  1.000
17-NOV-1983 / 23:  0.000  0.08333  1.000
17-DEC-1983 / 24:  0.000  0.08333  1.000
17-JAN-1984 / 25:  0.000  0.08333  1.000
16-FEB-1984 / 26:  0.000  0.04167  0.500
18-MAR-1984 / 27:  0.000  0.00000  0.000
17-APR-1984 / 28:  0.000  0.00000  0.000
18-MAY-1984 / 29:  0.000  0.00000  0.000
17-JUN-1984 / 30:  0.000  0.00000  0.000
 
 

Gary Strand wrote:

I looked at the user's guide and the FAQ, and couldn't quite find what I need.

Suppose I have a monthly average time series of a 1-D variable, (say) "TEMP".
If I create a 12-month running mean using "LET RUN = TEMP[L=@SBX:12]", how
can I list the weights used to generate RUN?

Thanks.
--
/\        Gary Strand (303) 497-1336                                NCAR CG2260
  \_][    www.cgd.ucar.edu/ccr/strandwg                    3080 Center Green Dr
      \___strandwg@ucar.edu                       Boulder, Colorado, USA  80301

--
Ansley Manke  Pacific Marine Environmental Laboratory  Seattle WA