Environmental Engineering Reference
In-Depth Information
2009a
,
2005a
,
2010
,
2007b
; Mopper and Schultz
1993
; Yamashita and Tanoue
2003a
; Komaki and Yabe
1982
; Nakajima
2006
; Baker
2005
; Chen et al.
2003
;
Burdige et al.
2004
; Fu et al.
2006
). Peak T
UV
-region depicts the shorter (UV)
wavelength ranges at Ex/Em
=
215-260/280-380 nm, which includes mostly the
secondary fluorescence peaks of various fluorescent organic substances such as
proteins, aromatic amino acids (tryptophan-like, tyrosine-like and phenylalanine-
like), algae, detergent component, phenol-like compounds, naphthalene,
o
-cresol,
p
-cresol,
p
-hydroxy benzaldehyde,
p
-hydroxy acetophenone, 1,4-dichlorobenzene,
and 4-biphenyl carboxaldehyde (Table
1
) (Mostofa et al.
2009a
,
2010
; Mopper and
Schultz
1993
; Yamashita and Tanoue
2003a
; Nakajima
2006
; Baker
2005
; Baker
and Curry
2004
; Chen et al.
2003
; Burdige et al.
2004
; Fu et al.
2006
). Note that
the fluorescence intensity expressed as QSU (quinine sulphate unit) is consid-
ered to identify the authentic or valid excitation-emission wavelength maxima.
In contrast, the maxima of the calibrated fluorescence intensity obtained using the
Raman Unit (RU: nm
−
1
) method can be significantly shifted, particularly at the
peak C-region (Mostofa et al.
2005b
). Such issues are discussed in details later, in
the fluorescence intensity normalization section.
2.3 PARAFAC Modeling in FDOM Study
Parallel factor (PARAFAC) modeling is a three-way multivariate analysis that can
be applied on an additive mixture of fluorescence signals obtained from excita-
tion-emission matrix spectra. PARAFAC is capable of isolating and quantifying
the individual fluorescence component signals in terms of fluorescence intensity of
FDOM in natural waters or in mixtures.
From the PARAFAC model (Harshman
1970
) it can be implied that for any
fluorophore, the emission intensity at a specific wavelength
j
that corresponds to
excitation at the wavelength
k
can be expressed as follows (Eq.
2.7
):
x
jk
=
ab
j
c
k
(2.7)
where
x
jk
is the intensity of the light at the emission wavelength
j
and excita-
tion wavelength
k
,
a
is the concentration (in arbitrary units) of the analyte,
b
j
is
the relative emission at the wavelength
j
, and
c
k
is the relative amount of light
absorbed at the excitation wavelength
k
. For any number of analytes and samples,
the PARAFAC model can be developed into a set of trilinear terms and a residual
array as (Eq.
2.8
) (Stedmon et al.
2003
)
F
x
ijk
=
a
ij
b
jf
c
kf
+ ε
ijk
,
i
=
1,
...
,
I
;
j
=
1,
...
,
J
;
k
=
1,
...
,
K
(2.8)
f
=
1
where
x
ijk
is the fluorescence intensity of the
i
ith sample at the emission wave-
length
j
and excitation wavelength
k
.
a
if
is directly proportional to the concentra-
tion (in arbitrary units) of the analyte
f
in sample
i
.
b
jf
is directly proportional to
