vertical cross section problem

[kbedka1]

I have attached an image from some cross section analysis I have been doing with a new product here. The magenta line in the lower left panel is the color filled contour cross section shown along the top. The arrow denotes a vertical profile that looks very strange. I used the vertical probe (green square, lower left) to find this profile and this is shown in the lower right panel. In matlab, I plotted all profiles within +/- 0.4 degrees of the lat/lon location of the vertical probe over the altitude layer with questionable data, see second attachment. Nowhere in this plot do I see this strange "zig-zag" profile behavior. It seems that Mc-V is fabricating data here. The color shaded cross section shows the same result, so it is not the contouring.

We counted on using Mc-V to analyze this case for an upcoming meeting in mid-March, but we cannot trust the results we're seeing here. Could someone look into this please? I'm using a nightly build from Feb 25th. My file is located at:

ftp://ftp.ssec.wisc.edu/pub/incoming/kris9.cdf

[Rick]

Hi Kris,

We are able to reproduce the problem when using nearest neighbor:

Edit -> Preferences -> Formats and Data

The only time we saw a similar graph is when we were right between 2 pixels (zoom way in and I think you'll see what we are talking about). As long as you stay within a pixel, you should be okay. I'll write it up as an inquiry. Nearest neighbor is the most scientifically correct way of depicting data, but we could not reproduce the problem using weighted average.

Rick

[tdrink]

Hi Kris,

Let me say first that McV will never fabricate information from data as only linear
based interpolation is ever used. I think Rick is correct, in that this is sort of an
aliasing problem. You can really see what's happening if you plot 2D plan views
of the Bill Cloud Water content, Bill Smith?, and examine the field by changing the
height of the 2D display around 5 to 6 km. In this case, weighted average might be better, you
could experiment a little. Weighted average use only a simple multi-linear
interpolation so it will only have a smoothing effect.

Tom

[kbedka1]

Thanks Tom, yes this is Bill Smith Jr's product. I have started using weighted average and this is working out better for me, no wierd artifacts. Have you been able to solve the 3-D volume rendering issue that I was having with this file?

[tdrink]

I'm trying a couple things right now.