[ncl-talk] ncl plot getting killed

Mary Haley haley at ucar.edu
Wed Nov 28 08:18:25 MST 2018


Hi Kim,

You could be correct that this is a memory issue. Are you getting any
errors from NCL before it quits?

You can save on memory by not creating so many local variables. Below are
four possible scenarios you can try:

[1] You can delete a variable after you no longer need it, which will free
up some memory.

no = aa[:]->no
no_avg=dim_avg_n(no,0)
delete(no)
no2 = aa[:]->no2
no2_avg=dim_avg_n(no2,0)
delete(no2)
oli = aa[:]->oli
oli_avg=dim_avg_n(oli,0)
delete(oli)

[2] You can use [/.../] to delete several variables at once, if you don't
need to free them individually:

no = aa[:]->no
no_avg=dim_avg_n(no,0)
no2 = aa[:]->no2
no2_avg=dim_avg_n(no2,0)
oli = aa[:]->oli
oli_avg=dim_avg_n(oli,0)
delete([/no,no2,oli/])

[3] Even better, if there's no reason that variables like "no", "no2", and
"oli" need to exist, then there's no need to save them to a local variable
before taking the average:

no_avg  = dim_avg_n(aa[:]->no,0)
no2_avg= dim_avg_n(aa[:]->no2,0)
oli_avg  = dim_avg_n(aa[:]->oli,0)

[4] Since you are doing the same operation repeatedly, you can use a do
loop and save yourself some more memory.

vars_to_avg = (/"no","no2","oli","csl","olt","xyl","ald","hc8",\
                         "hcho","ol2","tol","hc5","hc3","ket"/)
nvars = dimsizes(vars_to_avg)

;---Average the first one and store to variable
Mpan_avg = dim_avg_n(aa[:]->$vars_to_avg(0)$,0)

;---Loop thru rest of vars and sum their averages
do nv=1,nvars-1
  Mpan_avg = Mpan_avg + dim_avg_n(aa[:]->$vars_to_avg(nv)$,0)
end do
Mpan_avg=Mpan_avg*1000
Mpan_avg2D=Mpan_avg(0,:,:)
. . .

The above is not tested, so hopefully there are no syntax errors in it.

--Mary


On Wed, Nov 28, 2018 at 2:49 AM Komal Shukla <komalshukla1992 at gmail.com>
wrote:

> Hi
>
> I am trying to add a variable from netcdf files, and make an average plot.
> My scripts is mentioned below, but the process gets killed. Maybe because
> of memory.
>
> Can you please suggest a "efficient" way of writing this? Please help.
>
> Script ----------
> begin
> a = addfile("../wrfout_d01_2012-05-01_00:00:00.nc","r")
>
>  it        = 0     ; first time step
>   hgt       = wrf_user_getvar(a,"HGT_M",it)    ; Terrain elevation
>   hgt at lat2d = wrf_user_getvar(a,"XLAT",it)   ; latitude/longitude
>   hgt at lon2d = wrf_user_getvar(a,"XLONG",it)  ; required for plotting
>
>   wks = gsn_open_wks("png","try1")
>
>     FILES = systemfunc ("ls ../wrfout_d01_2012-05*") ; file paths
>
>  numFILES = dimsizes(FILES)
>  aa    = addfiles (FILES+".nc", "r")     ;add multiple files
> ;;;;;;;;;;;;
>
> no = aa[:]->no
> no_avg=dim_avg_n(no,0)
> no2 = aa[:]->no2
> no2_avg=dim_avg_n(no2,0)
> oli = aa[:]->oli
> oli_avg=dim_avg_n(oli,0)
> csl = aa[:]->csl
> csl_avg=dim_avg_n(csl,0)
> olt = aa[:]->olt
> olt_avg=dim_avg_n(olt,0)
> xyl = aa[:]->xyl
> xyl_avg=dim_avg_n(xyl,0)
> ald = aa[:]->ald
> ald_avg=dim_avg_n(ald,0)
> hc8 = aa[:]->hc8
> hc8_avg=dim_avg_n(hc8,0)
> hcho = aa[:]->hcho
> hcho_avg=dim_avg_n(hcho,0)
> ol2 = aa[:]->ol2
> ol2_avg=dim_avg_n(ol2,0)
> tol = aa[:]->tol
> tol_avg=dim_avg_n(tol,0)
> hc5= aa[:]->hc5
> hc5_avg=dim_avg_n(hc5,0)
> hc3= aa[:]->hc3
> hc3_avg=dim_avg_n(hc3,0)
> ket= aa[:]->ket
> ket_avg=dim_avg_n(ket,0)
>
>
> Mpan_avg=no_avg+no2_avg+oli_avg+csl_avg+olt_avg+xyl_avg+ald_avg+hc8_avg+hcho_avg+ol2_avg+tol_avg+hc5_avg+hc3_avg+ket_avg
> Mpan_avg=Mpan_avg*1000
> Mpan_avg2D=Mpan_avg(0,:,:)
> Mpan_avg2D at lat2d = wrf_user_getvar(a,"XLAT",it)   ; latitude/longitude
> Mpan_avg2D at lon2d = wrf_user_getvar(a,"XLONG",it)  ; required for plotting
> --some plotting options here--
> contour = gsn_csm_contour_map(wks,Mpan_avg2D,res)
> draw(contour)
>   frame(wks)
>
> Thanks
> Kim
> _______________________________________________
> 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/20181128/8d727561/attachment.html>


More information about the ncl-talk mailing list