An Innovative Approach to Normalize Soil Reflectance Spectra Acquired by Different Spectrometers and Operators
Eyal Ben Dor, Geography Department, Tel Aviv University
This paper was presented at the Art, Science and Applications of Reflectance Spectroscopy Symposium sponsored by ASD Inc. and IEEE GRSS, February 23-25, 2010 in Boulder, Colorado.
Authors: Eyal Ben Dor, Agustin Pimstein, Gila Notesco
Affiliation: Geography Department, Tel Aviv University, Tel Aviv Israel
Abstract
A wide range of factors affect the reflectance spectrum of a given soil sample. For example, variability can be introduced by the spectrometer itself, the experimental protocol and sometimes even by the operator. Unlike spectra of rock and minerals where easily recognized spectral features are sufficient for diagnostic assessment, it is more subtle spectral data that reveals the most characteristic information about a soil sample. Consequently, technical factors that alter the reflectance spectrum need to be identified and minimized and, if possible, eliminated or at least standardized. The need to minimize such factors in a robust way is important especially now that a worldwide soil spectral group has been established and soil spectral libraries are exchanged routinely between members. This paper summarizes a comprehensive and systematic study performed with this goal in mind. A group of 12 chemically defined soils were assessed spectrally by multiple operators using three different ASD spectrometers that require different protocols and measurement conditions. Despite the definition of strict protocols, for some soil samples there was great variation among the spectra generated by the three different ASD spectrometers that was exacerbated when more than one operator performed the spectroscopy. Therefore, we decided to investigate an approach taken from the wet chemistry discipline, whereby an internal and well-defined standard is included in every measurement cycle. Three different materials were examined for that purpose and a method was developed to normalize spectra to the spectra generated by a ‘mother’ spectrometer and protocol. We propose that this method enables the variable spectra of a given soil sample to be translated into a common denominator and thus should facilitate the exchange of spectral information among scientists worldwide and allow creation of a robust soil database for diverse applications.
Full article link