Re: JFM averages.

[Steve Hankin <hankin@xxxxxxxxxxxxx>]

Hi Marcelo,

I think that the axis you defined with
    define axis/edges/t="16-jan-1950:12":"16-dec-1994:12"/npoints=`45*4` tsea
has a problem. SHOW AXIS indicates a "drift" to your seasonal axis. Here we
examine every 5th year:

     yes? show axis/l=1:180:`4*5` tsea
     name       axis              # pts   start                end
     TSEA      TIME               180 r   03-MAR-1950 01:40
     31-OCT-1994 22:20
     T0 = 15-JAN-1901

                L     T                 BOX_SIZE       TIME_STEP (HOURS)
                1>  03-MAR-1950 01:40:00  2187.333       430657.7
               21>  27-FEB-1955 20:20:00  2187.333       474404.3
               41>  24-FEB-1960 15:00:00  2187.333       518151
               61>  20-FEB-1965 09:40:00  2187.333       561897.7
               81>  17-FEB-1970 04:20:00  2187.333       605644.3
              101>  13-FEB-1975 23:00:00  2187.333       649391
              121>  10-FEB-1980 17:40:00  2187.333       693137.7
              141>  06-FEB-1985 12:20:00  2187.333       736884.3
              161>  03-FEB-1990 07:00:00  2187.333       780631

Here's an off-the-cuff version of the axis -- using 15-FEB as a crude estimate
for the middle of the J-F_M season. Note there is no drift:

     yes? define axis/t="15-feb-1950":"15-feb-1995"/npoints=`45*4+1` tsea

     yes? show axis/l=1:180:`4*5` tsea
      name       axis              # pts   start                end
      TSEA      TIME               181 r   15-FEB-1950 00:00
     15-FEB-1995 00:00
     T0 = 15-JAN-1901

                L     T                 BOX_SIZE       TIME_STEP (HOURS)
                1>  15-FEB-1950 00:00:00  2191.467       430272
               21>  15-FEB-1955 05:20:00  2191.467       474101.3
               41>  15-FEB-1960 10:40:00  2191.467       517930.7
               61>  14-FEB-1965 16:00:00  2191.467       561760
               81>  14-FEB-1970 21:20:00  2191.467       605589.3
              101>  15-FEB-1975 02:40:00  2191.467       649418.7
              121>  15-FEB-1980 08:00:00  2191.467       693248
              141>  14-FEB-1985 13:20:00  2191.467       737077.3
              161>  14-FEB-1990 18:40:00  2191.467       780906.7

You can use the DAYS1900 function as shown in

http://ferret.wrc.noaa.gov/Ferret/Documentation/Users_Guide/v50/partII.html#V5.0_DEFINE AXIS$/EDGES

to get really detailed control over the axis definition.

    - steve

======================================

Marcelo Barreiro wrote:

> Hi ferreters,
>
> I am computing JFM averages from monthly data of a time series which spans
> 45 years (540 months).
> I did it in two different ways:
>
> (1st)
> define axis/edges/t="16-jan-1950:12":"16-dec-1994:12"/npoints=`45*4` tsea
> let my_var_seas = my_var[gt=tsea@ave]
> let/quiet my_var_JFM = my_var_seas[l=1:180:4]

>
> (2nd)
> let/quiet my_var_J = my_var[l=1:540:12]
> let/quiet my_var_F = my_var[l=2:540:12]
> let/quiet my_var_M = my_var[l=3:540:12]
> let/quiet my_var_JFM= (my_var_J + my_var_F + my_var_M)/3
>
> Results of my_var_JFM are similar but not equal and in cases the
> correlation between the results of different methods is as low as 0.90.
>
> Does anyone know which way is the best?
> Thanks!
>
> Marcelo
>
> ================================================
> Marcelo Barreiro
> Dept. of Oceanography, Physical Section
> College of Geosciences and Maritime Studies
> Texas A&M University
> College Station, TX 77843-3146
> Phone: (979) 845-8216
> Email: marcelo@ocean.tamu.edu
> URL: http://www-ocean.tamu.edu/~marcelo/
> ================================================

--
Steve Hankin
NOAA/PMEL, 7600 Sand Point Way NE, Seattle, WA 98115-0070
ph. (206) 526-6080 -- FAX (206) 526-6744