i1—25 November
,, 1984,
etation of
the northern
>82". JRC Report
, 1983,
it 4
i northern
ral Signatures
ir 1983,
iscence
ig 'Gelbstoff'
submitted to
, S.,A., 1984,
nguela
ebruary 1980."
eport on
in some
ing group
A 19620.
hop on remote
orthern
t JRC Ispra.
85.03 (1985).
n of satellite
the visible
etermination
yer of
elling spectral
. in
gy, EARSEL,
niv. of
., Ferrari ,G.,
nd total
nbus-7 CZCS".
1, Woods-
ii for sediment
nsitivity to
Y" •
an of Geodesy
983,
astal Zone
West African
e most
Africa,
Study"
rieval of
com remote
:al zone.
Symp. on
Louse
mtial of the
ispended
1C Internal
Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986
Shape and variability of the absorption spectrum of aquatic humus
H.Buiteveld* & F.de Jong
Delft University of Technology, Netherlands
* Present address: RijkswaterStaat, DBW/RIZA, Lelystad, Netherlands
R. Spanhoff
Rijkswaterstaat, DGW, The Hague, Netherlands
M.Donze
Kenia laboratories, Arnhem, Netherlands
ABSTRACT: Shape and variability of the absorption spectrum of aquatic humus is investigated. The exponential
description of the shape is not accurate enough for remote sensing applications. Considerable improvement in
the accuracy of the interpretation of airborne reflection measurements may be expected when actually measured
absorption spectra of aquatic humus (part of the optical seatruth) are used as input for the deconvolution
algorithm.
1 GENERAL INTRODUCTION
In numerous studies it has been demonstrated that
airborne passive remote sensing of surface water in
the optical window may yield a wealth of synoptical
information. This information consists of convoluted
data of the effect on the light-field of several
physical, chemical and biological compounds of
interest. Deconvolution is done using 'algorithms';
the results of these calculations are calibrated by a
statistical technique against some seatruth
measurements on the compounds of interest. The
calibrated values are subsequently used for
interpolation and extrapolation to construct
distribution maps of these compounds.
These maps are as yet of limited use due to the
noise and variability in space and time observed in
the 'constants' yielded by the calibration procedure.
This problem increases in importance going from the
open ocean to estuaries; in freshwater bodies it may
even be greater.
A number of factors may contribute to this
undesirable situation:
1. Lack of measurements; instrumental noise.
2. Natural noise in the environment.
3. Natural phenomena that are not recognized in the
measurements nor covered by theory.
4. Inadequacy of theory; especially nonlinearity in
the relationships between concentrations and optical
results cannot be treated in a statistically
satisfying way, given the amount of noise in the
data.
When we develop the instrumental and theoretical
apparatus to distinguish such factors we can reach a
position from which reliability of remote sensing
observations in dependence on the quality of
instrumentation, seatruth observations and local
conditions, can be judged.
As a first step it may be expected that 'optical
seatruth', consisting of optical measurements in situ
and spectroscopy of water samples, will be much more
usefull to develop precision in remote sensing than
attempts to directly calculate chemical
concentrations from airborne measurements. This
translation can be separately done with the
spectroscopic data.
Purpose of our research is to contribute to a
program as sketched above.
1.1 Aquatic humus
Humus (or yellow substance or gelbstoff) is a general
name covering dissolved organic compounds of large
molecular weight. Its definition actually consists in
the methods of isolation (primarily pore size of the
filterpaper) and measurement; such as total organic
compounds, extinction and fluorescence, or any more
elaborate set of properties.
Pure water has fixed optical properties. Humus has
a variable concentration and variable optical
properties (Zepp and Schlotzhauer 1981; Bricaud et
al. 1981). Together these two determine the optical
background in which the contribution from particulate
material must be studied. The optical properties of
water were reviewed by Smith and Baker (1981). In the
present paper the absorption spectrum of humus is
discussed.
The shape of the absorption spectrum of humus can
in first approximation be described by an exponential
function (Kalle 1966); absorption decreases strongly
with wavelength in a monotonous fashion.
a(A) = A e d(A_Ao) (i)
with a(A) = measured absorption coefficient in m 1
A = wavelength in nm
A 0 = arbitrary constant in nm
A,d = calculated by least squares
This exponential form is usually applied in marine
optics (Prieur and Sathyendranath, 1981). The
accuracy of this description was studied by Bricaud
et al. (1981) and Zepp and Schlotzhauer (1981).
In this model A may be roughly equated with the
concentration of humus, while d roughly describes the
shape of the spectrum. In fact both parameters depend
on the choice of Ao, the measured wavelength range and
individual deviation from the model. Zepp and
Schlotzhauer (1981) observed that values for A do
correlate with total organic carbon, with an
uncertainty of a factor of about 2.
Bricaud et al. (1981) determined the constant d
using a linear regression fit to equation 1 in the
range 375-500 nm. The value of d varied from -.02 to
-.01 nm 1 , with a mean of -.014 nm 1 . Zepp and
Schlotzhauer (1981) found, in the case of freshwater
humus, d values between -.0116 and .0175 nm 1 ,
with a mean value of -.0145 nm 1 , using the wavelength
range 300-500 nm.
The exponential function (1) with d = -.014 nm 1 is
often used as model for the humus absorption. But it
appears that the variability of the absorption
spectrum of humus in nature and the deviation from
the exponential function, with fixed d, are
considerable.
2 MATERIALS AND METHODS
Samples of surface water were collected at 10
different locations in The Netherlands.