nethods,
D and
1ded road
A Monte Carlo radiative transfer simulation
of rice canopy based on digital stereo photogrammetry
Keiji KUSHIDA and Kunihiko YOSHINO
Department of Agricultural Engineering
Faculty of Agriculture,
University of Tokyo
1-1-1 Yayoi, Bunkyo-ku, Tokyo 113, JAPAN
Phone: *81-3-3812-2111 (ext. 5345), Fax: *81-3-5684-3632
Commission VI, Working Group 1
KEY WORDS:
Vegetation, Radiation, Simulation, Agriculture, Matching,
Digital, Three-dimensional, Measurement
ABSTRACT:
NIR (Near Infra Red) bidirectional reflectance distributions of thirteen kinds of rice canopies are simulated
using the 3-D Monte Carlo method and measured in an experimental field. The simulation is carried out under two
different field conditions. One is similar to the condition of the experimental field (condition A) and the other is the
wide spread field condition under the parallel solar beam (condition B).
The framework of the simulation model is 1cm x 1cm x 1cm sized rectangular solid cell. Each cell has an
information on leaf area, inclination, and direction that are calculated using a 3-D digital stereo measurement
method. Other than the structural information, the sensitivity of individual leaf reflectance is accounted.
The sensitivity is decided by measuring some kinds of individual leaf reflectance. From the measurement,
NIR reflectance and transmittance of rice leaves have low coefficient of variation. On the other hand, VIS (Visible)
reflectance and transmittance have high coefficient of variation.
The simulated bidirectional reflectance fit the measured one well under condition A in NIR region. As the
result of the simulation, the ratio of the reflectance factors (view zenith: 45deg. to 0 deg. ) in NIR region from wide
spread field is suggested to give average inclination of rice canopy.
1. INTRODUCTION
Radiative transfer simulation of plant canopy is
basic to remote sensing of plant canopy state. A
complex scene is hard to model with analytical models.
With the simulation models such as Monte Carlo
method, one can consider canopy structure that is hard
to model with analytical models. Therefore, the
simulation has been carried out using the Monte Carlo
method.
Ross et al. (1988) modeled plant elements (leaf,
stem) as simple geometrical shapes. Round or elliptic
leaves are assumed and the difference of bidirectional
reflectance factor is estimated under the conditions of
different variance of leaf inclination, different deg. of
ellipticity, the row effect, and the influence of stems.
Kimes et al. (1982) represented information on
Plant canopy physical state as 3-D cell matrix and
developed a simulation model. A size of one cell is
about 10cm. Each cell that corresponds to leaf has
Information on leaf inclination distribution function and
leaf area index. Each cell has the same theoretical or
389
empirical leaf inclination distribution. Simulation is
carried out with homogeneous dense canopy (Kimes,
1984), homogeneous sparse canopy (Kimes et al.,
1985), and inhomogeneous canopy (Kimes et al.,
1986). The results of the simulation are compared with
measured value, and bidirectional reflectance
characteristics of various vegetation canopy are
explained.
‘However, simulations and validation of their
result have hardly carried out from the viewpoint of
estimate of canopy physical state of a fixed crop
canopy. One of the reasons is thought to be that the
structural measurement of canopy has been hardly
carried out so that the cells are small enough to
correspond to a minute region of an individual leaf.
G.V.Menzhulin et al. (1991) also pointed out this as a
shortcoming of the simulation model.
Therefore, NIR bidirectional reflectance
distribution of thirteen kinds of rice canopy is simulated
using the 3-D Monte Carlo method based on cell
information given from 3-D structure obtained by the
leaf edge matching method (Kushida et al, 1993;
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B7. Vienna 1996