I want to quantify the minerals from soil from hyperspectral images. I have collected the proximal ground samples. Now how to get the actual minerals composition from soil and how to get the mineral composition from image?
Dear Mr. Singh,
I do not know how to quantify the mineralogical composition of a soil or overburden in a 3-D system by means of a D-2 image processing, but I can tell you how you can quantify the mineral composition during ground-follow-up work.
1. Carry out mapping in the field in a 3-D and 2-D classical manner
2. Take samples not randomly but go on targeting upon reference profiles layer-by- layer
3. Carry out mineralogical work using the Rietveld analysis of XRD patterns at the finest way.
With kind regards
H.G.Dill
SEM-EDX based image analysis can work for identifying small to micro mineral grains, but it would be best if analyzed as a polished section. You can find “Mineral Liberation Analysis” (MLA) or “Qemscan” labs at several universities.
Hello Simranjit,
You don't say what hyperspectral imaging system you are using and its spatial resolution and spectral coverage. Judging from what is on the market it could be anything from 2mm to 30um and from VNIR-SWIR to LWIR. The larger the pixel size the more likely you will encounter spectral mixing, and even at the highest spatial resolution you will likely encounter intimate mixing. Furthermore, dusting of fine particles on coarse will determine what is visible to the sensor. You will also only detect minerals that are active in the spectral region you are sensing.
The hyperspectral images can be interpreted using end-member analysis and spectral unmixing, spectral matching to mineral libraries such as the USGS spectral library, or by extraction of diagnostic spectral parameters. Using training data from your ground samples you could produce a calibration of diagnostic spectral parameters against XRF, XRD or other mineral or chemical (normative mineral) analysis method.
As Mr Dill has explained, collection of 2-D images to attempt to quantify a 3-D system is limited; and variation of spectral features typically behave non-linearly to a multitude of factors as well as component abundance. Even with this type of calibration, your analysis of hyperspectral imagery will be, at best, semi-quantitative.
Kind regards
Brian
Dear Colleague,
The soil samples have been collected. XRD, reflected light and electron microscopy; and EMPA can be very helpful inqualitative and quantitave mineralogical composition determination of your soil samples.
My regards
Napoleon
For your soil analysis I recommend you first make a sample of soil, mesh type, after taking your sample you must homogenize it to be representative, then I recommend two good techniques:
Make transparent cuts for studies through optical microscopy of transmitted light, to see your minerals forming roa and alteration. You can also make polished cuts to see the opaque minerals in case you are near a hydrothermal system. There are transparent-polished cuts. Aca in Chile are not of good quality, so hagos separately.
The other to support everything seen in the microscope and if you do not recognize any mineral, you can do X-ray diffraction. Both methods that I recomend are economical and are robust data for mineralogy.
luck
- Badges
- Science topic
- Similar topics
- Computer Science
- Images