Multivariate calibration of chlorophyll a using partial least squares and electronic absorption spectroscopy

Abstract

Partial least squares (PLS) calibration has been applied in the study of electronic absorption spectra of pure chlorophyll a. Inverse univariate calibration was applied separately at each wavelength. The effectiveness of the calibration models was quantitatively expressed as root mean square errors of both calibration and prediction. Wavelengths with the best linear behaviour were selected, and PLS performed on an increasing number of wavelengths, sorted by their prediction efficiency. PLS calibration results in a substantial reduction in the prediction error relative to the best single wavelength result and is an alternative to using molar absorbance (extinction) coefficients.