| problem set [message #10717] |
Mon, 12 January 1998 00:00 |
Martin Schultz
Messages: 515
Registered: August 1997
|
Senior Member |
|
|
Dear newsgroup readers,
Here's an example for the trickiness of contour plots, and I
appreciate any help I can get to improve this thing:
First of all, it is almost impossible to describe what I see,
therefore I put the current version of the problematic plots on our
ftp server: ftp://www-as.harvard.edu/pub/exchange/mgs/idl_question.ps
It shows vertical profiles of 3 different atmospheric trace gases
vs. latitude and longitude, respectively. The color fields in the
background stem from a gridded POLYFILL output, and they are OK ...
... but it is hard to see a (scientific) message here. Therefore, I
would like to superimpose contours on this plot (shown as black lines).
Technically, this works fine. The trouble is the data, and the result:
(1) The contours are artificially closed near the boundaries of the
plot, producing vertical lines which distract the reader,
(2) The sparcity of data in some areas obviously produces some
"funny" looking results. In fact, if I use the MIN_VAL option in
CONTOUR, I only get very few lines.
I already used the coarser grid which is shown as POLYFILL rectangles
to smooth (average) the data, then resampling, triangulating and
trigridding it to produce the contour plot. Use of the /IRREGULAR
option in CONTOUR led to much worse looking results, and the /QUINTIC
smoothing of TRIGRID did not have any positive effects either.
I know that there may be data which is practically impossible to
visualize properly, and certainly aircraft data from atmospheric field
missions is a very good candidate for this: trace gas concentrations
often vary over orders of magnitude over the region, the flight track of
an airplane is only a line in a 3D volume, re-visiting the same place at
a different time may give completely different results. Yet, since the
tax payer has spent so much money on collecting the data in the first
place (and one can draw valuable conclusions from it, indeed), it would
be desirable to produce figures that show things nicely. So, if anyone
has experience with similarily ill-behaved data, please let me know what
you do.
Thanks,
Martin.
PS: David, if you tell me "That's all in my book", then I am (almost)
ready to buy it ;-)
------------------------------------------------------------ -------
Dr. Martin Schultz
Department for Earth&Planetary Sciences, Harvard University
186 Pierce Hall, 29 Oxford St., Cambridge, MA-02138, USA
phone: (617)-496-8318
fax : (617)-495-4551
e-mail: mgs@io.harvard.edu
IDL-homepage: http://www-as.harvard.edu/people/staff/mgs/idl/
------------------------------------------------------------ -------
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
