The pixels that result in those very high or low ranges are probably a few pixels that were negative in the FLAASH results. The results from FLAASH can sometimes produce negative pixel values where the pixels have very low radiance and do not model well. These are often areas of water or shadow. After FLAASH processing, you might mask the values below 0 to be 0. Then your NDVI values should look more reasonable.

Thak you.. MariM Before, I also have that kind of thought to mask it. But just not sure if that's appropriate. However, there is one thing that I forgot to ask you about callibration. I got an error massage after input band 7 to do callibration. An error massage is as following; tmcal_doit: An error has been occurred during processing Error: "Attempt to subscript Lmin with BANDS_PRESENT is out of range." The result may be invalid. PS. bythe way, i input Lmin and Lmax of band 7 to callibrate already. Thank you. Paula.

What version of ENVI are you using? If it is ENVI 4.8, then you should open your Landsat data as Tiff with metaata and you shouldn't have to process band 7 separately. There were some issues with Landat calibration of Geotiff files prior to ENVI 4.8.

Dear MariM, I use ENVI 4.6, not sure if that cause the problem. Basicly, I would like to do FLAASH and using band 7 to retrived the water.. Unfortunately, I can't callibrate band 7. So the NDVI value after I calculate band Math had some unusaul value. Thank you and looking forward for your reply. Paula.

It is a problem with that version for GeoTIFF data. You can calibrate band 7 by hand using Band Math if you like. The radiance calculation is fairly simple and in the on-line help for ENVI. However, having band 7 of Landsat is not enough to perform a water retrieval. You would need more bands (more spectral resolution) to perform water retrieval. You could do an aerosol retrieval though.

Thank you..MariM. I will try to do it as your suggestion. Paula.

Dear MiriM, I can't find the Band Math for band 7 calibration. Is that only one equeation or have to process in many step? If you can suggest me? Thank you so much. Paula.

The equation to convert to radiance in the help is not specific to the band, so you would use it for band 7 as well. The parameters such as LMIN, LMAX, etc are found in the metadata file (_mtl.txt). The help discusses how to use Band Math and write equations.