Some notes on the divergence and transformed divergence formulae in the remote sensing literature

Abstract

Divergence, and particularly Transformed Divergence, are key separability metrics in remote sensing, predominantly utilized for digital image classification to achieve thematic mapping. Despite the widespread use of Transformed Divergence, the mathematical expression of Divergence–essential for calculating Transformed Divergence–appears in the literature in two different formulations, leading to inconsistent results and confusion within the remote sensing community. This study addresses the original source of Divergence, clarifies its correct formulation, and examines implementations that calculate Divergence and/or Transformed Divergence based on the original formula. Additionally, it provides data to verify the accuracy of future implementations and quantitatively shows the erratic implications of incorrect calculations.