An Introduction to the Physical Basis for Deriving River Information by Optical Remote Sensing - Fluvial Remote Sensing for Science and Management - page 57

to large values of a w ( ) over this range. In the scattering-

dominated regime, radiance increases with depth due to
a greater number of scattering interactions in a thicker
water column. Conversely, radiance decreases with
depth in the absorption-dominated regime because more
photons are absorbed within a thicker water column. At
the transition between the two regimes is a wavelength for
which the radiance is equal for all depths; this transition
shifts toward longer wavelengths as suspended sediment
concentration increases. Remote sensing of shallow water
bathymetry and bottom type thus relies primarily upon
red and near-infrared wavelengths, whereas efforts to
infer concentrations of suspended sediment, chlorophyll,
or dissolved organic matter will primarily make use of the
blue and green portions of the spectrum. Note that these
results were based on radiative transfer simulations for
which pure water and suspended sediment were the only
optically significant constituents of the water column.
The presence of dissolved organic matter or other
substances can modify the optical properties of the water
column, for example. In general, all wavelengths are
always affected by both absorption and scattering to some
degree.

λ reflectance R B ( ), and the overlying water column, with

a volume reflectance R C (

λ

λ ). If the bed is brighter than

the water - that is, if R B ( ) - the bottom is

detectable as an increase in radiance relative to deep
water; shallower depths correspond to higher radiances.
This scenario is typical of clear water and highly reflective
substrates at red and near-infrared wavelengths, where
radiative transfer is dominated by pure water absorption.
The opposite case, where the bed is actually darker than
the overlying water column and R B (

λ ) > R C ( λ

), implies
that depth and radiance are inversely related; shallower
water appears darker than deeper water. In this case, the
presence of a relatively dark substrate at a small depth
truncates the water column and reduces the amount
of radiance scattered by suspended sediment or other
materials. This scenario occurs at shorter blue and green
wavelengths dominated by scattering, particularly when
suspended sediment concentrations are high. A third
possibility is R B (

λ ) < R C ( λ

), implying that the bed is
indistinct from the water itself, making the bottom effec-
tively invisible and precluding estimation of depth. For
the conditions examined by Legleiter et al. (2004, 2009),
this scenario occurred from 550-600 nm at the transi-
tion from scattering- to absorption-dominated radiative
transfer; at the wavelength of this crossover, radiance was
independent of depth.
Because bottom reflectance, along with water column
optical properties, determines the nature of the relation-
ship between depth and radiance, some knowledge of
different bottom types is useful for mapping bathymetry.
If the objective is to characterise the spatial distribution
of various substrates, knowledge of their spectral
characteristics is critical - these characteristics determine
the utility of remotely sensed data for distinguishing
among streambed features. Reflectance spectra for several
different bottom types measured along a gravel-bed river
via ground-based spectroscopy are shown in Figure 3.8a
(Legleiter et al., 2009). Certain substrates, such as
limestone bedrock, are much brighter than other, darker
materials, such as gravel. Moreover, some bottom types
have very distinctive spectral shapes, such as periphyton,
which is a type of algae that adheres to sediments
comprising the streambed. This spectral diversity is
conducive to substrate classification via remote sensing,
and the pronounced differences in overall brightness
and spectral shape among these three features imply that
limestone, gravel, and periphyton could be distinguished
from one another.
Substrate spectral variability could undermine efforts
to map bathymetry, however. Standard depth retrieval

λ ) ≈ R C ( λ

3.3.3 Reflectancepropertiesof thestreambed
andbanks

If the effects of the water column are somehow accounted
for (i.e., via an effective attenuation coefficient),
remote sensing techniques can be used to map channel
bathymetry and/or substrate composition. For either
of these applications, the reflectance properties of the
streambed are an important consideration because the
bottom-reflected radiance L B (

) depends on both depth
and bottom type. Even if water column optical properties
are uniform at the reach scale, the relationship between
depth d and L B (

λ

) can vary due to local variations
in the irradiance reflectance of the streambed, R B (

λ

).
Conversely, bathymetric variation can complicate efforts
to estimate R B (

λ

) and map different substrate types.
In other words, estimates of water depth ideally would
be informed by knowledge of substrate composition,
whereas mapping of bottom type would proceed based
on knowledge of depth. In practice, of course, neither
depth nor substrate type is known apriori . Inferring
either attribute from remotely sensed data thus represents
an under-determined problem, but spectral information
can provide additional leverage.
An important concept in this context is the bot-
tom contrast between the streambed, with an irradiance

λ