Re: [ferret_users] how to shade the significant correlations at 99%confidence level

[Ansley Manke <ansley.b.manke@xxxxxxxx>]

Hi Yun,
You can define variables based on your input variables to define the 
correlation -- see the script variance.jnl for ideas how to do this.  To 
list the contents of the script:

 yes? go/help variance

You will see that this script defines a variable for correlation, 
"correl". Here is an example of a shade plot between the two variables 
zonal wind and wind speed.  (Just illustrating here; I make no claims 
about  the validity of doing these statistics on 12 time points!)

  yes? set data coads_climatology
 yes? let p = vwnd
 yes? let q = wspd
 yes? go variance

 yes?   ! shade with a centered palette
 yes? shade/palette=light_centered/lev=(-1,1,0.2) correl

 yes? ! Or, to shade only the values you wish:
 yes? let/title="99% confidence level" conf_99 = if abs(correl) GT 0.31 
then correl
 yes? shade/palette=light_centered/lev=(-1,1,0.2) conf_99


Yun.Li@csiro.au wrote:

> Dear Ferret users,
>
> I am trying to plot a correlation map between SST (60N-60S, 30E-300E)
> and SOI of Australian winter (June-August) in 1950-2005.  
> Could you please let me know how to plot the correlation map with the
> shaded correlations significant at 99% confidence level (>0.31 or <
> -0.31)?
>
> Thanks,
>
> Yun Li
>
>