Re: [idvusers] Isosurface Interpolation

Thanks for the reply, Don and Stu!

Regarding the "discontinuities", I'll try to get an image for you - I'll describe them in the meantime.

There are several consistent areas in the vertical where we are seeing a quasi-stair-step in the isosurface. Your explanation of normalizing the radial azimuths may explain this. I'll install and test the 2.3 release and see if this helps any.

Is the normalizing of azimuths required, or is it intended for better efficiency? The reason I ask ties into my attempt to get the NSSL netCDF output into the IDV and whether I need to do some of this normalizing in the netCDF files or whether the IDV will take care of that for me.

While on this subject of the NSSL netCDF output, the way we have out data is Product/ElevationAngle/datafile.netcdf. So for Reflectivity, it would look like:


With regard to rendering a volume scan, can I give the IDV the top level directory and have it construct the volume from the subdirectory structure, or will I need to dump all the elevation angle data into one file?



Don Murray wrote:
Hi Kevin-

Kevin L. Manross wrote:
I have a couple questions regarding isosurfacing in the IDV. We are using it to look at isosurfaces of radar reflectivity and we're noticing several discontinuities in the vertical when viewing Level II data. Can anyone tell me what type of interpolation is being used for the isosurfacing? Is this technique something that can be modified via the IDV? If not, is it possible for a user (me) to write some sort of plugin to change this interpolation?

When you say discontinuities, what do you mean?  We just fixed a bug
in the 2.2 release where some of the levels were shifted and it made
for some weird isosurfaces.  Try out the 2.3 release and see if that
works better for you.

Since radar sweeps in a volume do not have a standard pattern for the
scans  (e.g. each sweep starts at a different azimuth and the number
of azimuths may vary by sweep), we normalize the sweeps to a 0-360 set
of azimuths, putting the closest radial to each azimuth in for the data.
This gives us a "rectified" domain which makes the isosurface alogrithm
work better (but maybe not perfect).  As to the details, it uses the
standard VisAD algorithm, the details of which I'm not sure of.  But
that could be answered by someone on the VisAD list.

Also, I've played around a little with the vertical scaling widget and it doesn't seem to have any appreciable effect. (I changed my vertical scale from 16000m to 20000m.) When viewing a radar domain, it would be great to have the vertical scale of a storm more proportional to its horizontal scale.

Stu's response is how I do it - use the Range and Bearing tool to
calculate the width of the box, set the  vertical aspect to 1 and
set the vertical range accordingly.

Don Murray                               UCAR Unidata Program
dmurray@xxxxxxxxxxxxxxxx                        P.O. Box 3000
(303) 497-8628                              Boulder, CO 80307

Kevin L. Manross           |  ** New Address **
CIMMS Research Associate   |     120 David L. Boren Bvd
NSSL : WRDD : SWAT         |     Rm 3923
<kevin.manross@xxxxxxxx>   |     405.325.6385 |
"My opinions are my own and not representative of
CIMMS, NSSL, NOAA or any affiliates"