[ncl-talk] Overwrite rstd in loop with regCoef

Dennis Shea shea at ucar.edu
Fri Feb 10 15:07:02 MST 2017


[1] Maybe

  coef = new((/nmon,nlat,nlon/),float)
  rstd  = new((/nmon,nlat,nlon/),float)
  do i =0,11
       coef(i,:,:) = regCoef(NAO(i,:),rTAS(i,:,:,:))
       rstd(i,:,:)   = onedtond(coef at rstd,(/nlat,mlon/))   ; make 2D
  end do


[2] If you have NCL 6.2.1 (9/2014), I'd recommend looking at
    http://www.ncl.ucar.edu/Document/Functions/Built-in/regCoef_n.shtml

; TAS is (/nmon,nyrs,nlat,nlon/)
; NAO is (/nmon,nyrs/)

  coef = new((/nmon,nlat,nlon/),float)
  rstd  = new((/nmon,nlat,nlon/),float)
  do i =0,11
       coef(i,:,:) = regCoef_n(NAO(i,:), TAS(i,:,:,:), 0, 0)
       rstd(i,:,:)   = onedtond(coef at rstd,(/nlat,mlon/))
  end do



On Fri, Feb 10, 2017 at 2:47 PM, Lynch, Cary D <cary.lynch at pnnl.gov> wrote:

> NCL-talk,
>
> I am using the function “regCoef” in a do-loop.  X is a 2 dimensional
> variable of size (/nmon,nyrs/).  Y is a 4 dimensional variable of size
> (/nmon,nyrs,nlat,nlon/).  The resulting regression coefficient is 3
> dimensional of size (/nmon,nlat,nlon/):
>
> ; TAS is (/nmon,nyrs,nlat,nlon/)
> ; NAO is (/nmon,nyrs/)
> ; nmon = 12; yrs = 30
>
> ; reorder TAS
>   rTAS = TAS(month|:,lat|:,lon|:,year|:)
>
>   coef = new((/nmon,nlat,nlon/),float)
>   do i =0,11
>   coef(i,:,:) = regCoef(NAO(i,:),rTAS(i,:,:,:))
>   end do
>
>
> For each “i” I want to archive the *rstd* (standard error of the
> estimated regression coefficient), which will be in 1 dimension (nlat*
> nlon).
>
> Using the the special "$” syntax in the loop, doesn’t work.  Is there
> some other special syntax I could use?
>
> Thank you
>
> _______________________________________________
> ncl-talk mailing list
> ncl-talk at ucar.edu
> 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/20170210/10f843fa/attachment.html 


More information about the ncl-talk mailing list