The development of imaging spectrometers heralds a revolution in our ability to use remote sensing in the earth sciences. The combined high spectral and spatial resolution of these new sensors should motivate a shift in interpretation techniques towards more deterministic and quantitative methods. Geophysical inversion theory provides the proper framework for the development of these new approaches to remote sensing interpretation. Models relating the observable radiance to the controlling parameters are inverted along with auxiliary data and error models to arrive at quantitative estimates and uncertainties of the properties of interest. Linear spectral unmixing is a simple but useful example of one of the many inversion procedures that can be envisioned for these new remote sensing data. 1.
Joseph W. Boardman,
"Inversion of high spectral resolution data", Proc. SPIE 1298, Imaging Spectroscopy of the Terrestrial Environment, (1 September 1990); doi: 10.1117/12.21355; https://doi.org/10.1117/12.21355