CRISM DATA PROCESSING: CRISM Ratioed I/F Corrected Spectra

[Chandan Kumar]

Hi,

I need some help to understand the data processing of CRISM data (TRR3). I use CAT tool coupled with ENVI 5.3. I use CAT manual (workshop manual available in the website) for the processing which goes as Conversion of PDS to CAT--> Photographic and Atmospheric Correction--> Georeferencing. 

I observe that after the atmosphere correction spectra do not change much but once perform the geometric correction it shows changes as shown in the figure. Is obtained result correct? What is the main reason for this change?

I would also like to know how to obtain ratioed spectra. I am familiar with spectra ratios but my confusion is how to choose spectra (denominator) which does not introduce errors, what region' pixels would be a better choice? Is atmospherically corrected spectrum same as I/F spectrum? 

All plots are atmospheric corrected.

Thanks,

[Ray Arvidson]

Hi There. You are doing the processing correctly. The reason that the map projected plot looks different relative to the before projected plot is that the map projected plot has its scale enlarged. It does look like these plots have been made using Classic ENVI and CAT rather than the ENVI5.3 interface and CAT. I so I recommend you switch to the ENVI5.3 interface, much more powerful than the Classic interface.

Ray Arvidson

CRISM Science Team Member and PDS Geosciences Node Manager

[Ray Arvidson]

Oh and to remove artifacts in the spectra many scientists ratio the data to a "bland" spectrum from the same column in the data. You just need to make sure that the spectrum you ratio to does not have spectral features that would make some artifacts in the ratioed data. I personally do not do the ratios for this very reason.

Ray Arvidson

[Chandan Kumar]

Thanks a lot, Dr. Ray Arvidson for your reply and suggestion. I have used the classic version of ENVI & CAT for the processing.  

[Harish]

On 7/19/2018 at 3:42 AM, Ray Arvidson said:

Oh and to remove artifacts in the spectra many scientists ratio the data to a "bland" spectrum from the same column in the data. You just need to make sure that the spectrum you ratio to does not have spectral features that would make some artifacts in the ratioed data. I personally do not do the ratios for this very reason.

Ray Arvidson

Sir I am also facing the same problem. Actually I am not getting any neutral spectra in the whole column and image, so my query is - is it necessary to do rationing?

Also continuum removed spectra is showing clear adsorptions. So can I go with the continuum removed spectra?

[Ray Arvidson]

Yes, I tend to use continuum removed spectra rather than ratioed spectra because the former does not introduce any biases introduced by ratioing to non-gray spectra. You might first subset your data to the wavelength interval containing the candidate absorptions and shoulders in order to highlight just the absorption features of interest.

Ray Arvidson

[Harish]

On 12/13/2018 at 10:25 PM, Ray Arvidson said:

Yes, I tend to use continuum removed spectra rather than ratioed spectra because the former does not introduce any biases introduced by ratioing to non-gray spectra. You might first subset your data to the wavelength interval containing the candidate absorptions and shoulders in order to highlight just the absorption features of interest.

Ray Arvidson

Thank you sir. I will follow paper entitled ''Smectite deposits in Marathon Valley, Endeavour Crater, Mars, identified using CRISM hyperspectral reflectance data'' in this regard.