cleanFilter works on full image only

Errors and unexpected results
Post Reply
User avatar
joleenf
Posts: 1123
Joined: Mon Jan 19, 2009 7:16 pm

cleanFilter works on full image only

Post by joleenf »

Hi,

I am trying to use the clean filter on the visible image from the GOES-14 SRSO on 2014/05/08 at 16:48 UTC. It seems that when I tak a subset of that image at full res (896, 2000) centered at 43.4, -95.1, the cleanFilter does not work. It does work if I use all of the data at full res. Not sure why that happens. Could you check this? (This is through the gui formulas, not attempting the scripting api).

Joleen
User avatar
jayh
Posts: 424
Joined: Thu Jan 15, 2009 10:34 pm

Re: cleanFilter works on full image only

Post by jayh »

Hi Joleen-

I'm noticing on the 16:48 image from 2014/05/08, that neither the subset or the full res display with the cleanFilter is cleaning up the bad line through Iowa & southern Wisconsin. What I did notice was that when doing the full res display with filter is that the initial display looks cleaned up, but if you zoom in a little bit, you'll still see the bad line.

I checked with -X, and IMGFILT does not clean up that line with the default settings for FILT=CLEAN. (full res or subset). I needed to change the ldiff parameter in the CLEAN filter (a value less than 9 worked for me).

However, when I tried this approach on the GOES14 image directly from the server, I could not get cleanFilter to remove the line, no matter what ldiff value I tried. So, I tried a different approach. I copied the image with IMGCOPY in -X, then made a local dataset pointing to the AREA file. This worked.

I'm concerned it's an issue with using an image filter on a real time GVAR server, or perhaps an issue with the archive server. I wrote up inquiry 1744 to take a look at this: http://mcidas.ssec.wisc.edu/inquiry-v/?inquiry=1744

Let us know if you have further questions.
Thanks, Jay
User avatar
jayh
Posts: 424
Joined: Thu Jan 15, 2009 10:34 pm

Re: cleanFilter works on full image only

Post by jayh »

Hi Joleen-

I wanted to give you an update on some further testing I've done with this image. We are still in the process of analyzing the behavior of the SRSO GOES-14 imagery with IMGFILT when trying to clean up the ropey lines. TimS has given us some further images to test with as well.

Concerning your 2014/05/08 16:48Z image, I'm now seeing consistent behavior when loading data from the archive server vs. a local adde dataset (area file). Here's my summary:
- Loaded the images straight from the datasets with no filter to confirm presence of bad line over IA and WI, Band 1 - Brightness - Full Resolution
- Zoomed in on the bad line and set Auto Projection off.
- Loaded the images with the cleanFilter defaults, the line was not cleaned on either the remote or local data source.
- Tried ldiff=10... this cleaned the line on the remote data source AND local data source.
- Tried ldiff=3... this cleaned the line on the remote data source AND local data source.

So, I'm pleased that the archive server and local data source are working the same. I also confirmed the IMGFILT behavior in -X (CLEAN filter defaults don't clean up the line, ldiff of 10 or less do clean up the line) is the same in cleanFilter in -V.

I'm going to update 1744 with this information. We are seeing some inconsistencies with an example that TimS has given us, so there still may be further action in this inquiry to make sure IMGFILT is working similarly to cleanFilter.

Thanks, Jay
Post Reply