Rochdi Posted November 16, 2020 Posted November 16, 2020 (edited) Hello Experts, I would like to learn how to calculate some indices to detect Gypsum and other minerals on Ritchey crater, I decided to start first with the formulations included in "CRISM Data Product Software Interface Specification" (as seen below) to seek out gypsum. However, I didn't clearly understand the syntax of the mentioned formulations, I mean (R, a, b ... etc), I found these formulations more detailed compared to the indices I've calculated before such as NDVI, EVI, SAVI ... (using Landsat8, Spot imageries). Would you please explain to me the syntax of these formulations and what do I need to get started with before diving into this process? Look forward to your reply, Rochdi. Edited November 16, 2020 by Rochdi
Jennifer Ward Posted November 16, 2020 Posted November 16, 2020 Rochdi, Please see the attached paper for an explanation of how the indices are calculated. Jennifer Viviano-Beck_et_al-2014-Journal_of_Geophysical_Research__Planets.pdf
Rochdi Posted November 17, 2020 Author Posted November 17, 2020 Hi Jennifer, Thanks a lot for the attached paper, I found it useful for my Master thesis I'm working on. However, the mentioned indices above are still unclear to me for putting them into action. What I'm wondering about is these questions: Does R2210 mean reflectance at 2210nm? Do I need a value to assign it to R2210? if yes, could I use ENVI and CAT to get R2210 value, for example? I look forward to your reply, Rochdi.
Ray Arvidson Posted November 17, 2020 Posted November 17, 2020 Yes, for each pixel in the scene R2210 means use the reflectance value at 2210 nm. ENVI's Band Math function will allow you to write an equation that when run produces the parameter map.The tool is in the ENVI5 series interfaces on the right hand side under Band Algebra. Also, the weightings a, b, etc. in the equations allow centering of the computation over the deepest part of the putative absorption feature, all relative to the wavelengths on the continuum on either side of the feature. Ray Arvidson
Rochdi Posted November 17, 2020 Author Posted November 17, 2020 (edited) Hello Sir Ray Arvidson, Thanks a lot, that's what I thought before. I used the ENVI tool you mentioned, I tried to write the formulation of HCPINDEX just to learn. HCPINDEX: (R1470 - R1050/R1470 + R1050) * (R1470 - R2067/R1470 + R2067) As I wrote the expression, I refered R1470, R1050 and R2067 to b1, b2 and b3 successively (X1.png). Then, a new window popped up to assign the variables (b1, b2 and b3) to their specific wavelengths. The following is the bands that I assigned to the variables (X2.png): b1 (R1470nm) => Band 73 (1473, 7nm) b2 (R1050nm) => Band 09 (1053.8nm) b3 (R2067nm) => Band 255 (2067.4nm) However, I come up with this unexpected output result as seen below (X3.png) Is something went wrong? Edited November 17, 2020 by Rochdi
Ray Arvidson Posted November 17, 2020 Posted November 17, 2020 Your band math seems to be missing parentheses at the appropriate places. Please read the band math documentation for how to represent algebraic expressions. Ray Arvidson
Rochdi Posted November 17, 2020 Author Posted November 17, 2020 Yes, after I read the documentation for how to represent algebraic expression on ENVI Band Math function. I realized that I need to modify the equation by adding float in the dominator as follows: HCPINDEX: ((float(b1) - float(b2)) / (float(b1) + float(b2))) * ((float(b1) - float(b3)) / (float(b1) + float(b3))) I come with this attached output result below. So, where's Pyroxene, then? or the result should be in three color compsites?
Ray Arvidson Posted November 17, 2020 Posted November 17, 2020 Result is one monochrome image. Do your band math in sensor space, use CAT to flatten the result, and then project. Results will have to be your task. I do not know where the pyroxene is located. Ray Arvidson
Rochdi Posted November 17, 2020 Author Posted November 17, 2020 So, is the result I got above is the expected one?
Ray Arvidson Posted November 17, 2020 Posted November 17, 2020 It should be a monochrome image with bright areas indicative of a high value for the spectral parameter used, yes. Ray Arvidson
Rochdi Posted November 17, 2020 Author Posted November 17, 2020 Alright, then. Concerning other indices which have different reflectance values compared to L cube bands values. In this case, Can I assign the bands which are closest to the reflectance values of those indices? Rochdi.
Ray Arvidson Posted November 18, 2020 Posted November 18, 2020 Yes, although the Viviano-Beck paper that we sent to you shows which bands to use for her indices. If you use your own indices and select accordingly. Ray Arvidson
Rochdi Posted November 18, 2020 Author Posted November 18, 2020 That's it, Sir Hope I'm not bothering you, as you know the L (IR) cube is already converted into IOF. If I want to extract the spectral signature of these bright areas which highlighted in the output above, could I use CAT toolkit for this purpose? or just a tool from ENVI toolbox?
Ray Arvidson Posted November 19, 2020 Posted November 19, 2020 Rochdi: I am not sure what you are asking. CAT has spectral parameters for IR data. Use this function. I am not trying to be judgmental. However, I suggest that you need to read in detail the Viviano-Beck paper that was sent to you and that you become very familiar with CAT and its functions. Many of your queries would have been answered if you had done this homework in detail. Ray Arvidson
Rochdi Posted November 20, 2020 Author Posted November 20, 2020 Hello Sir Ray Arvidson, Sometimes I ask stupid questions, but I extracted the spectral signatures of those bright areas. Thanks a lot for your tremendous help Sir Ray, and thanks to Jennifer G. Ward for the mentioned paper. Rochdi.
Ray Arvidson Posted November 20, 2020 Posted November 20, 2020 Rochdi: Glad to have helped you along. Stay safe and healthy. Ray Arvidson
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now