I'm not sure that I understand your question. Are you asking how does ENVI know how many bands are in your input image? Or are you asking why there are so many spectral measurements for a single spectrum in one of the spectral libraries that comes with the ENVI installation? Those libraries contain spectra collected with laboratory spectra, which measure hundreds of wavelengths. When you use them as endmembers for a SAM classification of an image, ENVI should resample them to match your image. Most of the spectral libraries that come with ENVI don't have spectra that continue out into the thermal bands covered by ASTER. I am guessing that that is why you are seeing an "invalid" message somewhere. You would need to find spectra that do cover the thermal wavelengths represented by the ASTER thermal bands to use as endmember spectra for the SAM classification. For example, spectra in the jhu_lib libraries continue into thermal bands. Or you could use spectra from pixels in the image itself, that you know contain the material you wish to map using Spectral Angle Mapper. - Peg Exelis VIS

While collecting the endmembers for SAM, in the endmember collection dialogue box, when I am importing an ASCII file using the option "from ASCII file", it is showing 6 columns in the dialogue box after importing the spectrum naming "Spectrum name","Color", "Source", "Bands", "Wavelength", "Status". In the "Bands" column, there are showing 923 bands. Where does this come from? And in the "Status" column it is showing "invalid". Why? If I select one particular mineral from the jhu_lib, the no. of bands are showing as 2287 & the status is still "invalid". What should I do to extract only the bands from 8µm to 12µm which is generally supposed to be the thermal bands region for ASTER imagery?

What is the source of your ascii file? It is from a spectrometer? How many bands do you expect your input spectra from ascii to have?

I checked the ASTER spectral library from Arizona State University spectral library which is available online. From there I saved the spectra as an ASCII file. The option for saving the spectra as an ASCII file was provided there. That ASU spectral library is most probably made up of laboratory spectra with the help of spectrometer. http://speclib.asu.edu/libmaker.php As I am working with the ASTER thermal bands only, I expect there to be only 5 bands instead of so many bands showing there.

If the data is from a spectrometer, then it likely has hundreds of bands so it sounds like ENVI is importing it correctly. A spectral library will be a series of columns with the first column representing the wavelength, the second may be FWHM (may or may not be present), and the subsequent columns will be the reflectance value. It is reasonable to use a spectral library with many bands in a classification with an image that has fewer bands. The spectra, if the units are known for both the image and spectral library, will be resampled on the fly and applied appropriately.

In such a case, I understand that there are so many bands. For example, ASTER band 10 ranges from 8.125-8.475 µm. In the imported ASCII file spectra, for a single mineral the above said range is further divided in narrower bands. For SAM classification how can I utilize all the values of each narrow band for a single thermal band?

If you import the spectra to use in SAM for an image such as ASTER, the spectra will be automatically resampled to match the range of ASTER.