[ncl-talk] escorc: Non-fatal conditions encountered in series or xstd equals zero.

Dennis Shea shea at ucar.edu
Sun Mar 8 13:15:11 MDT 2015


Did you look at the definition of cross-correlation used by escorc

cor = SUM [(X(t)-Xave)*(Y(t)-Yave)}]/(Xstd*Ystd)
note ... division by standard deviation(s)

==

The message says that one or more grid point series has all constant values.

If all are 0.0 or 10 or whatever, the standard deviation is zero.

Division by 0 is undefined so ... NCL warns you of this.
NCL calculates the cross correlation at all other points.
==

significance of r: min=7.031559566397816e-07   max=0.9999383172328334

The significance of all the grid points span 0 (no correlation) to
0.999 (*100=99.9%)



On Sun, Mar 8, 2015 at 9:17 AM, Vanúcia Schumacher <
vanucia-schumacher at hotmail.com> wrote:

> Dear all,
>
> I'm trying cross-correlation and the significance rtest, but I am getting
> the following error when calculating the weather and anomaly data:
>
> warning:escorc: Non-fatal conditions encountered in series or xstd equals
> zero.
> Possibly, all values of a series are constant.
> warning:escorc: Most likely, one or more series consisted of all constant
> values
>
> Another problem that I see are the values of very strange rtest output!
> I do not know if there's something wrong.
>
> Variable: rsig
> Type: double
> Total Size: 129600 bytes
>             16200 values
> Number of Dimensions: 2
> Dimensions and sizes: [lat | 90] x [lon | 180]
> Coordinates:
>             lat: [ -89..  89]
>             lon: [   0.. 358]
> Number Of Attributes: 2
>   long_name : significance of r
>   _FillValue : 1.000000020040877e+20
> (0) significance of r: min=7.031559566397816e-07   max=0.9999383172328334
>
> The last part of the script (attached) does not understand that the
> correct would plot (output > 0.05 or 0.95 to 95%)
> ShadeGtContour(plot2, 0.95, 17) or ShadeGtContour(plot2, 0.05, 17)
>
> Thanks,
>
> _______________________________________________
> ncl-talk mailing list
> List instructions, subscriber options, unsubscribe:
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ucar.edu/pipermail/ncl-talk/attachments/20150308/ce377cc8/attachment.html 


More information about the ncl-talk mailing list