Hi-
You could put the data onto a 1-D grid, sort it, and then find the
middle value. That would go something like this (I have not tested
these commands, but you'll get the idea.)
let var1d = xsequence(analysed_sst)
let isort = sorti(var1d )
let varsort = samplei(var1d, isort)
let npts = varsort [i=@ngd]
let median_value = varsort[i=`INT(npts/2)`]
Ansley
On 9/13/2011 10:40 AM, francoise orain wrote:
Hello
I need a median (Quantile0.5) like average on a time series (in a
concatenate file netcdf ) lat lon day
- I succeed for mean I use average AVG
use analyse.nc
name title I J
K L
ANALYSED_SST
analysed sea surface temperatur 1:1251 1:601
... 1:30
let MOY=ANALYSED_SST[L=0:30@AVE]
shade MOY
let MEDIAN=ANALYSED_SST[L=0:30@MED]
shade/l=6 MEDIAN (because need a 2D field)
I wanted to do the same thing with @MED but I realize that it is
not a
median (statistic sense ) but a smoother on one time
I shoud want a real median on the period for every points like
average
AVG (gif joined)
- My question is : is there a way to have this median or
quantile
90 with ferret for a 3D file or I have to use an another
graphical
tool
thanks a lot
Françoise Orain
|