You can divide the data by 10000 using Band Math.

I divided the image by 10,000. The output was a range 0 to 3. What I expected was 0-1. What may be the reason. I think in any case it should not be beyond 1.

It is possible that the FLAASH correction wasn't perfect, and resulted in a few pixels with reflectance values greater than 1 (and/or less than 0). This isn't physically possible, but it is mathematically possible, and does sometimes happen during FLAASH corrections. If there are only a few pixels with these physically unreasonable values, it shouldn't be much of a problem. If there are a lot of pixels that have reflectance values greater than 3 (which means they had values greater than 30,000 before you divided by 10,000), then your FLAASH correction may have had some problem with the input parameters, or may just not not be very good for this particular situation. Or perhaps the FLAASH output was not really scaled to have reflectance between 0 and 10,000. - Peg

Specular reflection can cause reflectance values greater than 1. If there are only a few pixels with reflectance > 1 and they are predominantly smooth and highly reflective objects (mirrored glass, certain kinds of arirplane fuselage, car windshields, etc) I wouldn't worry. Also, FLAASH multiplies reflectance values (whatever those might be) by 10,000, it doesn't scale them from 0 - 10,000.

Thanks a lot Peg. By changing one of the input parameter, we could get satisfactory result, still few pixels are having vauge values but thats okay. Is it okay to validate the FLAASH results with that of field spectra with the gap of one year(i.e. satellite date of pass and ground data collection).

Thanks a lot sallux, only few of the pixels are having vauge values, so we can proceed as per your suggestion.