While many types of spectroscopic selectivity problems are of
an additive nature (e.g. absorbance effects of chemical interferents),
others have a strong multiplicative components. Examples
of the latter are light scattering variation or optical path length
variations. If these effects change uncontrollably from sample to
sample, it is advantageous to reduce their effect in the preprocessing
step. Otherwise their multiplicative nature may otherwise destroy
the subsequent additive PLSR calibration modeling. If the
physical and chemical effects in the spectra are adequately different,
multivariate statistical modeling may be used for their separations.
One method is multiplicative signal correction (MSC) [24]
which searches for correcting the baseline and amplification effects
to the same average level in every spectrum. MSC was developed
into the extended multiplicative signal correction (EMSC) by Martens
et al. [24,25] in 1985 to get a more effective separation of
chemical and physical effects in light spectroscopy. The EMSC
method applies knowledge about the spectra of the analytes and
interference effects to improve the path length estimation.