CRISM Bad Bands

[Rachel Mastrapa]

I first wanted to say how impressed I have been by all of the support material and software for CRISM. I really appreciate all of the hard work that everyone has done, especially as a new user.

 

My question regards the bad bands list. I have been following the Nilli Fossae walkthrough while processing a set of data from Ganges Mensa. The processing works well until the step where I clean the spectral cubes. The first time that I tried processing a cube, the software failed. I determined that there were some bands that had a lot of values of 65535 throughout the image and not just on the left side. By editing the bad bands list to exclude these bands, I can get the despising and destriping to run. I have noticed that I have to make sure to get rid of every "bad" band because, if there are a couple still in there, the clean can produce ringing in the spectrum. However, since I am using the NIR from bands 4-254, I have started trimming off the rest to save space and time.

 

My questions is this: does trimming off the wavelength affect the destriping and despiking? Do I have to use the entire wavelength range for it to work properly, or does it not matter? I have been editing the header in the original CAT file so that all of the corrections are on the same bands. Please let me know if change the wavelength range changes any of the other programs as well.

 

Thanks in advance for your help! I can provide images and examples of what I am talking about if needed.

 

Best Wishes,

Rachel Mastrapa

[Susie Slavney]

Rachel,

I have forwarded your question to a software developer on the CRISM Team, Frank Morgan.

Susie Slavney

[Susie Slavney]

Rachel,

The short answer to your question is no, it doesn't matter if you have trimmed the wavelength range.

I asked some CRISM team members and that was the consensus, after going through all their responses.

In particular, Frank Morgan says,

 

"There's one other step she might have to take to get the CIRRUS routines to

work on TRR3 data.

 

In addition to the spectral subsetting Rachel does, she might also have to

remove the two lines at each end of the image. Those are filled with 65535

at all bands in TRR3 data and can cause a "No good data..." failure.

 

With ENVI Basic Tools/Resize Data, click "Spatial Subset" and in the "Lines"

boxes, make sure the minimum is >= 3, and the max is <= NL-2 (where NL is

the number of lines in the full image, usually something like 420, 480,

etc).

 

If Rachel's already doing spatial subsetting and removing the lines at the

ends, this won't be a problem."

 

Hope this helps. If you have more questions, post them here; I'll see them.

 

We all apologize for taking so long to respond.

 

Susie