Radiance and reflectance are completely different things, and they do not have the same units.  The radiance used as input to FLAASH needs to be in units of µW/cm2 * nm* sr.  Or if it is in some other radiance units, then you will need to specify a scale factor that, when divided into the input image, results in units of µW/cm2 * nm* sr. Reflectance is a ratio of incoming radiance to outgoing radiance, and is therefore unitless.  Reflectance values range from 0.0 (no refletance), to 1.0 (all incoming radiation is reflected).  The reflectance that is output from FLAASH is, by default, scaled to range between 0 and 10,000.  This is done so that it can be stored in a two-byte integer, rather than in a 4-byte floating point value.  If you want to compare the spectra in reflectance images output from FLAASH, you may want to use ENVI's Spectral Analyst tool (under Spectral > Spectral Analyst).  In this tool, you specify a spectral library, a method(s) of comparison, and you can also specify x and y scale factors to account for differences in the x (wavelength) or y (reflectance) scaling. It is not uncommon to find some negative reflectance values in FLAASH output.  It means that FLAASH overcorrected those pixels.  Usually this overcorrection is slight.  If the negative values are small, and relatively rare in the scene, then it may be reasonable to interpret them as pixels with zero reflectance at the band that contained the negative FLAASH output. Peg

Great! Peg, Thank you. Tutorial's FLASSH output image seems like to have the reflectance bound(-3*10^4 ~ 3*10^4). So, can I remove the bound(-3*10^4 ~0 , 10^4 ~3*10^4) by considering them with zero reflectance? And, in ENVI, can I get scaled spectral library reflectance(0 to 10,000) as same scaled of FLASHH output reflectance? Or, can I get scaled FLAASH output reflectance (0 to 1)?