IEEE Geoscience and Remote Sensing Magazine - June 2019 - 14

Specifically, we varied the horizontal visibility (the surface meteorological-range VIS parameter in MODTRAN)
in Figure 4(b); the WV scaling factor (H2OSTR in MODTRAN) in Figure 4(c); the sensor viewing angle i v (ANGLE
in MODTRAN) in Figure 4(d); shadowing (by means of
a sh) in Figure 4(e); topography (by taking z - z sl = 0 and
varying i sl) in Figure 4(f); adjacency effects (using a nullalbedo or olive rangeland or aluminum signature for the
adjacent material) in Figure  4(g); and obstacles and obstruction (using a null-albedo or olive rangeland or aluminum signature for both 15% and 30% sky-dome obstructions) in Figure 4(h).
With respect to the basic configuration of VIS = 23 km
(the typical default value in MODTRAN for good visibility
conditions), visibility variations strongly affect the spectra,
as shown in Figure 4(b). The major effects can be observed
in the low-wavelength region for low VIS values, where the
low target reflectivity combined with the increased scattering phenomena due to the hazier conditions results in considerable increases of the at-sensor spectral radiance. For
higher wavelengths, where scattering not only is generally
weaker but the copper target also has higher reflectance,
the general effect on the at-sensor radiance is a decrease of
the signal for low VIS values due to the less transparent atmosphere (lower transmittance). The result is a completely
distorted signal. In contrast, in Figure 4(c), the effects of
WV can be seen to be mostly confined to the WV absorption bands, with decreased signal (more absorption) for
more humid atmospheres (higher WV scaling). Oblique
sensing [Figure 4(d)] plays a significant role only for very
slanted views (e.g., i v = 120c), which mostly affect lower
wavelengths with an increased at-sensor radiance due to
the increased scattering phenomena occurring in the longer

target-to-sensor atmospheric path. Very oblique views also
affect, although to a lesser extent, the signal in the major
absorption bands (e.g., at 940 nm), where decreased signal
intensities can be observed due to the higher absorption.
The reduction of the direct solar illumination due to shadowing entails considerable variability over the at-sensor
spectral radiance in Figure 4(e), with a general decrease of
the at-sensor signal throughout the VNIR range. Similar
variations can be observed in Figure 4(f), where varying the
inclination of the targeted surface determines a general increase (if the surface is rotated toward the sun) or decrease
(if it is rotated away from the sun) of the at-sensor signal. In
the simulated scenario, the sun zenith angle is i = i i . 31c,
and thus an inclination of i sl = 30c is close to the condition
of maximum illumination due to topography.
The adjacency effects shown in Figure 4(g) illustrate
how much those effects depend on the properties of the
adjacent material itself. In this example, the low-reflectivity
rangeland spectrum hardly affects the resulting at-sensor
radiance. Conversely, for adjacent materials with stronger
reflection properties, such as the olive or the aluminum materials, adjacent effects are more significant. The olive signature "colors" the at-sensor radiance according to its own
spectral characteristics, with adjacency effects mostly manifesting themselves in the NIR spectral range; in the VIS,
these effects are negligible, because the olive material and
copper have a null contrast in that range of the spectrum.
The aluminum object, which has a higher reflectivity than
copper throughout the spectral range, causes a general increase in the at-sensor signal. Similar considerations can be
found for the secondary illumination due to nearby objects
[see Figure 4(h)], which cannot be considered negligible
for high-reflectivity objects. Here, the shape factor induces

TABLE 2. A CATEGORIZATION OF ENVIRONMENTALLY INDUCED SOURCES OF VARIABILITY.

ADJACENT
ENVIRONMENT

ACQUISITION
GEOMETRY

SOURCE OF
VARIABILITY

MAINLY
AFFECTED TERMS

SPATIAL
STATIONARITY

TEMPORAL
STATIONARITY

Constituents,
gases, aerosols

Ed (m), Lp (m),
Td (m), Tu (m), S(m)

Stationary over
wide areas*

Seasonal

Clouds, WV

Td (m), Tu (m),
Lp (m), Ed (m), S(m)

Nonstationary

Nonstationary

Adjacent materials

Ladj (m), ra (m)

Nonstationary

Stationary

Nearby objects
and obstacles

Ldif (m),

Nonstationary

Stationary

Shadows

Ldir (m)

Nonstationary

Hourly variations

Sensor viewing
angle

Tu (m), Lp (m)

Stationary**

Stationary

Sun position

Esu (m), Ed (m),
Lobst (m), Ladj (m),
Td (m), Tu (m)

Stationary over
wide areas

Hourly variations

Topography

Esu (m), Ldif (m),
Lobst (m), Ladj (m)

Nonstationary

Stationary

Lobst (m)

ATMOSPHERE

ILLUMINATION

*Except for wide-imaged areas and situations with extremely wide-angle views or very shallow slant-angled views.
**Except for situations with extremely wide-angle views.

14

IEEE GEOSCIENCE AND REMOTE SENSING MAGAZINE

JUNE 2019
IEEE Geoscience and Remote Sensing Magazine - June 2019

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2019
