BI-DIRECTIONAL REFLECTANCE DISTRIBUTION FUNCTION (BRDF) RETRIEVALS
FROM LABORATORY MULTIANGLE MEASUREMENTS
Alessandro Barducci*, Donatella Guzzi, Paolo Marcoionni, Ivan Pippi
*National Research Council — Institute for Applied Physics “Nello Carrara”, via Panciatichi 64, 50127 FIRENZE, ITALY -
Submit to: "TS ThS 2 Multi-angular Sensors"
KEYWORDS: Bi-Directional Reflectance Distribution Function, multiangle observation, hyperspectral instrument, data processing,
parameter retrievals.
ABSTRACT:
Theoretical investigations and intense experimental activity have being devoted to study the geometrical properties of reflection from
a surface, that is expressed in terms of Bi-Directional Reflectance Distribution Function, a quantity that is relevant to many remote
sensing applications, such as atmospheric modelling and climatologic studies. A large amount of laboratory and in-field
measurements are now available from various systems such as the European Goniometer Facility of the Joint Research Centre, the
Field Goniometer System of the University of Zurich, and the Portable Apparatus for Rapid Acquisition of Bidirectional Observation
of Land and Atmosphere instrument of NASA — GSFC. Moreover, recent satellite missions such as the Multi-angle Imaging
SpectroRadiometer and the Compact High Resolution Imaging Spectrometer on board of EOS-AMI and PROBA platforms supply
experimental data to this research. Following this general trend aimed to improve the current understanding of directional properties
of reflection from a surface, we present laboratory multiangular observations of natural sands. These data are used in conjunction of
literature data to study the reflectance dependence on surface optical and mechanical properties, illumination and viewing geometry.
A simple empirical model is developed in order to retrieve Bi-Directional Reflectance Factor from experimental data. Finally, the
possibility to extend the proposed physical model even to remotely sensed data is discussed, with particular reference to images
gathered with CHRIS spectrometer.
1. INTRODUCTION
The knowledge of absorption of solar radiation by canopy and
soil is of great interest to atmospheric modellers and
climatologists, since it determines to a large extent the amount
of solar energy effectively available for the whole climate
system (Reichman, 1973). This absorption of radiative energy is
also of concern to agronomists and biologists because it directly
affects the physiology and productivity of plants (Chappelle et
al, 1992),
While the amount of radiation actually absorbed by a natural
target is difficult to measure, it is more practical to retrieve the
absorption coefficient as a residual {from the measurement of the
surface and volume scattered radiation.
However, many natural surfaces exhibit preferential directions
for the reflection of solar irradiance. The measured reflectance
of such a surface depends not only on its structure and the
position of illumination source but also on the relative position
of the observer. This is the major inconvenience to estimate
“conical-hemispherical” reflectance (albedo) a(A,9.,¢,) of the
surface, since the scattered radiation has to be measured over
different viewing directions (Pinty and Ramond, 1986):
2 (%,90.94,9,,9, )c0sd dQ,
2n,
aQ.,9,.0,)- (D)
fro, ‚6; ) cos9,dQ ,
Qj
where the denominator indicates the total irradiance impinging
on the concerned target at angle (85,065). L,(,94,04,9,,0,)
the reflected radiance collected by the sensor in the direction
(9,,0,).
Spectral reflectance is generally retrieved considering the
observed target as a Lambertian and homogeneous diffuser for
which the upward radiance can explicitly be expressed as a
function of surface reflectance (Barducci and Pippi, 1994; Qiu,
2001).
Although many surfaces behave similarity to an ideal diffuser,
the above assumption will fail in two areas. The first obvious
failure is that many flat surfaces have a not negligible specular
component (for instance oceanic water in the visible spectral
range and high-reflecting soil), i.e. an increase in the observed
reflectance when the illumination and viewing zenith angles are
the same and the relative azimuth angle is 180°. Many rough
surfaces (like vegetation canopies) also show a reflectance
increase in the “hot spot” direction when the illumination and
viewing zenith angles are the same but the relative azimuth
angle is null. Moreover, the error resulting from assuming
Lambertian reflection for a natural target can be large for off-
nadir views in remote sensing observations (Martonchik et al.,
1998; Miesch et al., 2002.).
Theoretical investigations have being devoted to study the
geometrical properties of reflection from a surface, that is
expressed in terms of its Bi-Directional Reflectance Distribution
Function (BRDF), a quantity that takes into account both the
illumination and viewing geometry. The BRDF is a theoretical
concept that describes the directional reflectance by relating the
incident irradiance from a given direction to its contribution to
the reflected radiance in another specific direction.
