I've found that it is very easy to leave errors behind in the rasp.run.parameters.REGIONXYX file. The errors usually have to do with not starting a run at the same time as there is a grib file, or trying to generate forecasts after the last grib file. These leave maddeningly obscure error traces, obviously because they are stupid errors that should not have been made...

Has anyone worked out a table/chart of what is permissible (and reasonable given forecast spin-up), as far as time of run and time of output? We know that the "standard" ones work, but what else is OK?

These two scripts are checks to run after you have created a new domain to make sure all the well-known but subtle issues have been take care of.

Multiple Nested Domains under the same Parent -- No rerun of parent.

Post processing diagnosis, rerunning failed plots, and avoid most race conditions

These are probably idiosyncratic to my setup where I have a not-so-reliable connection to my web server.

Site management scripts for blipspots and windgram sites

The result is a "poor man's database" for your sounding/windgram sites.
New site data are simply added to the bottom of an ascii data file with this format
sitedata.ncl

DOMAIN,Sitename,grid,lat,lon,

1. A rewrite of the data for use in bash scripts that invoke extract_blipspots.PL
2. A rewrite of the data for into a javascript soundings menu. (I have adapted my univiewer so that the menu of soundings and windgrams is generated by a separate javascript.)
3. A rewrite of the data into two separate javascript menus one for all blipspots and one for all windgrams. These get incorporated into other pages on my site.
4. The creation of two simple html pages for accessing blipspots and windgrams
5. An output of a new rasp.ncl.region.data to put sounding labels on plots
6. Finally a new set of
   push@{$PARAMETER_DOLIST{'FRASER'}},('sounding1','sounding2','sounding3','sounding4','sounding5','sounding6','sounding7' )
   lines for each region.

• Most of the above are sorted from North to South by having a temporary lat at the beginning of the output line.
• Lines for different purposes are identified with keywords so that they can be pulled out of the overall output using grep.
• There is an additional script makestaticsoundings that makes static pages for soundings for people with slow connections.

Warning Note:

rasp.ncl changes

Warning: all of the scripts below require some modification to work on your site.

My rasp.site_load.xxxxx.ncl scripts.

Additional color maps I have worked out for use with RASP

Windgrams and other plotting scripts

Warning: all of the scripts below require some modification to work on your site.

• The output is now png by default and can be controlled by settin the type variable. The output is much crisper, but the file size is very much larger.
• Added an optional pressure contour to show whether heights are rising/falling or staying the same over the course of the day.
• Has flags for turning the available "markers" on or off in the plot.

  ; options to display or not on the windgram  
  show_lapse = 1          ;local level to level lapse rate ( ΔC°/Δft  expressed as C°/1000ft)
  show_wind  = 1          ; windbarbs 
  show_rh    = 1          ; relative humidity greater than rh_cut when above your cut level faded white cross hatches will be drawn.
  show_temp  = 1          ; temperature contours
  tmp_units = 1           ; change to 0 for C for Celcius 1 for Farenheight --  only affects the show_temp variable.
  show_up_vel= 1          ; show vertical velocity W* at top of time column
  ; polymarkers all drawn on lapse_plot, so lapse plot must be set to 1.
  show_snow  = 1          ; lowest freezing level
  show_wing  = 1          ; hg/pg 225ft/min 1.2m/s   soaring level markers
  show_lcl   = 1          ; little clouds for lifted condensation level
  show_p_contour = 1      ; constant pressure lines rise/fall steady during day.
  show_z_contour = 1  	; constant altitude  to the compare to the pressure contour 
  
  Other setable parameters include 
  debug = 0   ; anything other than zero prints debug info
  useblipdata = 0 ; use pre calculated data from the param.curr.hh00lst.grid.data blipdata files.
  unit_scheme = 0 ; still waiting for someone to properly implement this unit_scheme
  show_condense = 0 ; show level where temp==dewpoint probably useless rh above seems to work reasonably well.
  show_rain = 0  ; activate Andrea Barcelona's rain code
  show_pbl =  0  ; polymarker for boundary layer +show_pbl 
  hide_lapse_colors = 0  ; hide_lapse_colors -- used if you want color windbarbs, and lcl, but no lapse colors
  
  type = "ncgm"  ;legitimate values are x11, ps, ncgm, png 
  ptop = 24      ;top of the plot in pressure levels.  Change to suit your audience and region.
  rhcut = 95     ;plot crosshatching where rh > rhcut -- where there are likely to be clouds 
  DOMAIN = "DOMAINXYZ"
  SITE = "SITE"
  
  Any or all of the above can be passed on the command line if so desired. 
  But the string variables must be passed enclosed in excaped double quotes.
  e.g.  ncl windgramtj.ncl type=\"png\"
  
  In addition, there are default parameters within the program that you should change to suit your setup.
  outdir="$BASEDIR/RASP/UTIL/windgrams/"    ; could be anywhere
  operator="TJ Olney"                       ; your name here :-)  credit to me is always nice though
  default_domain= "PNW-WINDOW"                      ;your default domain here
  default_grid="w2"
  default_lat= 48.61
  default_lon= -122.43
  default_type = "x11"
  default_ptop = 30                        ;usually 20 I changed for TS upper instability
  default_rhcut = 95
  
  
  ;color choices  must be in the palette being used or the closest one in the current pallette will be used.
  foreground_color= "white"
  background_color= (/.5, .5, .9 /)  deep purple-blue
  label_color= foreground_color      
  wstar_color= "yellow"
  line_color= 13
  templine_color= 14
  colorpalette = ""
  
  
  

• Andrea Barcelona's code for rain is now included and working for me. It required some tweaking to run on both d2 and w2 domains.
• A datestamp for when the plot is actually plotted.
• changed absolute paths to $BASEDIR paths -- less work for you.
• The "updraft" colormap is included inline so there is no need to install it. (Some people don't like the colors...)
• Temperature as a potential variable to plot as a contour plot, on and Farenheight by default.

1. Download windgramtj.ncl and sitedata.ncl to your ncl directory
2. Ensure that you are using NCL v 5.1 or later! (or be prepared to rewrite the data ingestion routines that rely on addfiles.)
   NCL version 6.0 seems to be fully compatible with windgramtj.ncl
3. Modify sitedata.ncl to match your regionxyz and names of sites and lat and lon of sites within your regions.
4. Create a line at the end of $BASEDIR/RASP/RUN/run.rasp to run the windgram script after all times for a given regionxyz have run.
5. Changes you should consider making to windgramtj.ncl
   1. Modify the outdir to suit your setup, or ensure that the outdir exists
   2. Optional: Change units back to metric if you so desire by removing the *3.2808, *196.85, and *1.94386 conversion factors.
      Note that if you do, you will also want to change

          uv_res@vcLevels                 = (/ 3.,6.,9.,12.,15.,18.,21. /)   ;knots
          uv_res@vcLevelColors            = (/2,3,4,5,6,7,8,9,10/)
      

      to whatever units you want for wind speed. And lapse rates from this

      lapse_res@cnLevels =            (/ -3.,-2.5,-2.0,-1.5,-1.2, -.5, 0.,.5 /)
      lapse_res@cnFillColors         = (/ 11,10,7,8,3,-1,-1,13,14 /)         ; -1 is transparent
      

      To your unit-equivalents where -3C°/1000ft is absolutely unstable 0 and greater is inverted.
      Point of interest here is that here, near the Canadian border, most Canadians still think of height in feet, not meters but speed is kph.
      
   3. Choose your default values for ptop, rhcut, and SITE. ptop is a pressure level index, not an actual pressure.
   4. Comment out, remove, and/or revise as needed the 5 or six-line sections labeled "kludge" that are idiosyncratic to my site.
   5. Choose which "markers" to plot. Default is LCL as calculated by NCL which differs from any of DrJack's various lcl measures and freezing level — which I have found that many long time pilots use as their intuitive surrogate for lapse rate.
   6. Test it from the command line:
      ncl windgramtj.ncl [ptop=xx] [SITE=\"sitename\" | DOMAIN=\"REGIONXYZ\" | \"all\"] [rhcut=xx] [type=\"x11 |ncgm |ps \"]
   7. To make windgrams for your site
      The windgrams made by NCL are now png files but are not particularly suited for putting on the web because of filesize.. You will need to use the same method that the rest of the RASP system uses to convert these to .png or .jpg or .gif files. You can do this automatically through the last line of run.rasp, or using a cron job that runs the windgramtj.ncl script after all the times are in for a region. The same job should convert the *.ncgm files to the desired graphic format, usually .png, move the images to a web space, and do any archiving or house cleaning that you desire.
      See my example script getwindgramtj for the syntax to use with convert.

      PNG notes.

      Since NCL 51.1, png can be output by default. In the first implementations of this, the resulting images were wonderfully crisp, but the file sizes were huge. They also take a long time to render, making them not suitable for cgi windgrams or for doing all the RASP plotting. For a small finite number of windgrams, I use png output. I have used a method to shrink them that involves decreasing the color depth in two stages. for file in $yyyy_mm_dd*.png do /usr/bin/convert $file -depth 4 temp.gif /usr/bin/convert temp.gif $file done

I am currently testing the cloudwater variables, recently discussed on the forum, to plot instead of rh>95. If it works, I'll post when working. October 2009 -- Have studk to rh>95.

To Do: