How the band names are stored in the metadata DIM and how they are arranged in the file are different for SPOT. SPOT stores the TIFF as if was going to be displayed as a CIR with band 1 = NIR, band 2 = red and band 3 = green. However, the band names are XS1, XS2, etc. and are stored in order in the DIM. These values should be applied to the correct bands when calibrating. Why do you believe the result from Radiometric calibration is wrong?

The original value of Near infrared band= 81 and green band = 109. After I used the METADATA.DIM as input for the Radiometric Calibration in ENVI5.0, the resulted value for XS3 = 64.40598 (physical gain XS3=1.692295,physical offset =0) and XS1 = 40.68635 (physical gain XS1=1.990840, physical offset =0). While I did manually using the digital-to-radiance fomular: [radiance = (digital number/calibrationgain) + calibration offset], the result for XS3= 47.86399 and XS1= 54.75075. I think the result using radiometric calibration in ENVI should be the same with the manual one. I wonder that whether my understanding is correct or not? Additionally,my finally purpose is to atmospheric correction the original SPOT file. So, I use the file with radiometric calibration as input for running FLAASH atmospheric correction.