20001003: SAVS() function

James,

I used the cursor garea command in gdmap to define an area of the eta211 grid
with3x3 points showing....and actually, the number of grid rows and
columns in my input area was 5x5 points  (shown by gdlist). The graphical
X display will clip the numbers whose display won't fit in the full graphics 
area.

Looking at the $GEMPAKHOME/source/gemlib/df/dfsavs.f routine, the
subset area is that defined in DGCMN.CMN, and from that I found that
the COMMON  / DGAREA / value of kextnd is 3, so the computation
extends 3 rows/columns on all sides of the 5x5 array, or the
effective calculation is 11x11 (121) points.

I believe that the extended region is used for 2 reasons. First,
the values needed for some computational quantities (dirivatives)
need adjacent rows/columns. The second reason is that the
contouring up to the edge of the graphics region would probably
look very poor if the computation did not extend.

So, that is the explanation of how the value is arrived at. I do not have 
a simple suggestion of how to turn off the extended rows in the savs computation
if you only want the values of the internal grid points other than to compute 
the
value from the gdlist output. 

Steve Chiswell
Unidata User Support. 

>From: "James R. Kirk" <address@hidden>
>Organization: UCAR/Unidata
>Keywords: 200010031349.e93Dn7b12779

>Hi Steve,
>   Thanks for the reply.  I am a professor at Edinboro Univ. but also a
>graduate student at Penn State.  I TELNET to a Sun Workstation at Penn State
>named "stokes" and run GEMPAK using x-windows from my PC in my office at
>Edinboro.  I am not sure what version of GEMPAK I am using but I would
>imagine that Penn State is one of your sites.  If the version is critical,
>then I can find out.
>
>    I have tried using GAREA to set the grid subset.  It does not seem to
>work.  For instance, I plotted "vor(wnd)" and used CURSOR GAREA to define a
>certain region.  The region contains only about nine grid point values (as
>plotted in GDMAP).  The nine values of vorticity in the region are all
>around 25 X 10^-6.  When I change GFUNC to "SAVS(vor(wnd))" the nine values
>all change to -1.22 X 10^? (I forget the power).  It does not seem to be
>averaging just the subset, but the whole grid, although when I use SAVG, my
>recollection is that I get a different result.  I am not sure what it being
>averaged, but it certainly is not the nine grid points.  Any advice?  Thanks
>in advance.
>
>Jim Kirk
>(address@hidden)
>
>-----Original Message-----
>From: Unidata Support <address@hidden>
>To: James R. Kirk <address@hidden>
>Cc: address@hidden <address@hidden>
>Date: Monday, October 02, 2000 5:56 PM
>Subject: 20001002: Fw: GEMPAK Question
>
>
>>
>>James,
>>
>>I don't recognize your location as one of our sites. It would help us
>>out to know some more information about you, including how you obtained
>>GEMPAK, your GEMPAK version, patch level, the OS you are using etc.
>>along with the relevant input you are using to set the parameter values
>>in your program.
>>
>>The GAREA should define the subset of points used for the calculation.
>>GAREA can be defined several ways, shown in the "phelp garea" online
>>help information.
>>
>>
>>Steve Chiswell
>>Unidata User Support
>>
>
>