Re: time series plotting

As suggested by Verica .... polymarkers: something like
 res_at_xyMarkLineModes = "Markers" ; choose which have markers
 res_at_xyMarkers = 16 ; choose type of marker
 res_at_xyMarkerColor = "red" ; Marker color

----
If you prefer connected lines, perhaps the following will work.
Let "time" and "y" by the original variables. The
following assumes that the time steps are exactly 30 sec apart.
   ntim  = dimsizes(time)
   tStrt = time(0)
   tLast = time(ntim-1)   ; or could be specified
   dt   = 1800        ; seconds (30 min)
   NTIM = (tLast-tStrt)/dt + 1
   TIME = fspan(tStrt, tLast, NTIM)
   Y    = new ( NTIM, typeof(y) )   ; set to default _FillValue 
   
   do n=0,NTIM-1
      i = ind(time.eq.TIME(n))
      if (.not.ismissing(i)) then
           Y(n) = y(i)
      end if 
     ;delete(i)        ; not necessary here
   end do
   plt = gsn_xy(wks, TIME, Y, res)
good luck
On Wed, 10 Jan 2007, Verica Savic-Jovcic wrote:
> Hi James,
> 
> Why don't you just plot in Markers mode instead of Line mode?
> 
> Verica
> 
> ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
> http://www.atmos.ucla.edu/~verica/
> 
> ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~
> "The art of being wise is the art of knowing what to overlook."
>                                                William James
> 
> 
> On Jan 10, 2007, at 4:59 PM, James McPhee wrote:
> 
> >Hi All,
> >I'm trying to plot a time series with data gaps. This is I have something
> >like
> >
> >X!0="time"
> >X&time=stamp
> >stamp_at_units="seconds since 1-1-1970 00:00:0.0"
> >
> >Most of the data is taken every 30 min, but some times are not available. The
> >unavailable times do not correspond to _FillValue values; instead, the time
> >corrdinate just doesn't include those time stamps. Therefore, if I use
> >gsn_csm_xy to plot the data, a straight line joins the last point before a
> >gap and the first point after the gap. I'd like the plot to actually show the
> >gap, for which I imagine I would have to somehow *insert* the missing time
> >stamps and corresponding _FillValue values into the time series. Is there an
> >efficient way to do this in NCL without going through a loop? can I avoid
> >inserting the missing time stamps at all and still have the desired effect on
> >the plot?
> >
> >Thanks much!
> >
> >James
> >
> >James McPhee
> >Profesor Asistente, Departamento de Ingenieria Civil
> >Facultad de Ciencias Fisicas y Matematicas, Universidad de Chile
> >Av. Blanco Encalada 2002 piso 3
> >Santiago, Chile
> >tel: +56-2-9784400
> >jmcphee_at_ing.uchile.cl
> >http://www.cec.uchile.cl/~jmcphee
> >
> >
> >
> >_______________________________________________
> >ncl-talk mailing list
> >ncl-talk_at_ucar.edu
> >http://mailman.ucar.edu/mailman/listinfo/ncl-talk
> 
> _______________________________________________
> ncl-talk mailing list
> ncl-talk_at_ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/ncl-talk
> 
_______________________________________________
ncl-talk mailing list
ncl-talk_at_ucar.edu
http://mailman.ucar.edu/mailman/listinfo/ncl-talk