I am looking at the exact reference that the ENVI docs give for these tools, which is: J.A. Richards, 1999, Remote Sensing Digital Image Analysis, Springer-Verlag, Berlin, p. 240+. In this reference, on p. 245 (with a figure on p. 244) it explicitly states that the JM distance is "asymptotic to 2.0 so that a JM distance of 2.0 between spectral classes would imply classification of pixel data into those classes with 100% accuracy." This reference on p. 246 also specifically shows the transformed divergence ranging from 0 to 2.0, with 2.0 also corresponding to a 100% classification accuracy. I hope this helps. - Peg Exelis VIS

Thanks

What is the difference between Transformed Divergence (TDV) and Jeffries distance (JMD)? . I cannot understand that for between some classes i get lower value of JMD (around 1.45) but value of TDV is around 1.9. but still both of them gives statistical seperability of classes. I want to explain some of my results with this spectral seperability. Is it fine if I use any one of them ? Thanks