In: Wagner W., Székely, B. (eds.): ISPRS TC VII Symposium - 100 Years ISPRS, Vienna, Austria, July 5-7, 2010, IAPRS, Vol. XXXVIII, Part 7B
171
RETRIEVAL OF VEGETATION BIOCHEMICALS USING A
RADIATIVE TRANSFER MODEL AND HYPERSPECTRAL DATA
R. Darvishzadeh a ’ *, Clement Atzberger b , Andrew Skidmore c , Martin Schlerf c
a RS & GIS Department, Faculty of Earth Sciences, Shahid Beheshti University, Tehran, Iran- (r_darvish@sbu.ac.ir)
b Joint Research Centre of the European Commission, TP 266, Via Enrico Fermi 1, 21027 Ispra (VA), Italy -
(clement. atzberger@j re. it)
c NRS Department, ITC Faculty, University of Twente, Enschede, The Netherlands - (skidmore@itc.nl;schlerf@itc.nl)
Commission VII
KEY WORDS: Hyper spectral, Vegetation, Estimation, Model, Spectral
ABSTRACT:
Accurate quantitative estimation of vegetation biochemical characteristics is necessary for a large variety of agricultural and
ecological applications. The advent of hyperspectral remote sensing has offered possibilities for measuring specific vegetation
variables that were difficult to measure using conventional multi-spectral sensors. In this study, the potential of biophysical
modelling to predict leaf and canopy chlorophyll contents in a heterogeneous grassland is investigated. The well-known PROSAIL
model was inverted with HyMap measurements by means of a look-up table (LUT). HyMap images along with simultaneous in situ
measurements of chlorophyll content were acquired over a National Park. We tested the impact of using multiple solutions and
spectral sub-setting on parameter retrieval. To assess the performance of the model inversion, the RMSE and R 2 between
independent in situ measurements and estimated parameters were used. The results of the study demonstrated that inversion of the
PROSAIL model yield higher accuracies for Canopy chlorophyll content, in comparison to Leaf chlorophyll content (R 2 =0.84,
RMSE=0.24). Further a careful selection of spectral subset, which comprised the development of a new method to subset the spectral
data, proved to contain sufficient information for a successful model inversion. Consequently, it increased the estimation accuracy of
investigated parameters (R 2 =0.87, RMSE=0.22). Our results confirm the potential of model inversion for estimating vegetation
biochemical parameters using hyperspectral measurements.
1. INTRODUCTION
The spatial and temporal distribution of vegetation biochemical
and biophysical variables are important inputs into models
quantifying the exchange of energy and matter between the land
surface and the atmosphere. Among the many vegetation
characteristics, leaf chlorophyll content (LCC) and canopy
chlorophyll content (CCC) are of prime importance. Leaf
chlorophyll content and canopy chlorophyll content (the latter
defined here as the product of LAI and leaf chlorophyll content)
contribute to verifying vegetation physiological status and
health, and have been found useful for detecting vegetation
stress, photosynthetic capacity, and productivity (Boegh et al.,
2002; Carter, 1994).
The physical approach for estimating vegetation parameters
from remotely sensed data, involves using radiative transfer
models. This approach assumes that the radiative transfer model
accurately describes the spectral variation of canopy reflectance,
as a function of canopy, leaf and soil background
characteristics, using physical laws (Goel, 1989; Meroni et al.,
2004). As radiative transfer models are able to explain the
transfer and interaction of radiation inside the canopy based on
physical laws, they offer an explicit connection between the
vegetation biophysical and biochemical variables and the
canopy reflectance (Houborg et al., 2007). To actually use
physically based models for retrieving vegetation characteristics
from observed reflectance data, they must be inverted (Kimes et
al., 1998). A drawback in using physically based models is the
ill-posed nature of model inversion (Atzberger, 2004; Combal
et al., 2002), meaning that the inverse solution is not always
unique as various combinations of canopy parameters may yield
almost similar spectra (Weiss and Baret, 1999). To overcome
this problem, some restriction of the inverse problem may be
required to constrain the inversion process. This involves the
use of prior knowledge about model parameters (Combal et al.,
2002; Lavergne et al., 2007).
Significant efforts to estimate and quantify vegetation properties
using radiative transfer models have been carried out in the last
two decades. Despite these efforts, literature reveals that studies
on heterogeneous grasslands with combinations of different
grass species and the use of hyperspectral measurements are
rare. The main objective of this paper is to estimate and predict
canopy and leaf chlorophyll content by inverting the canopy
radiative transfer model PROSAIL (Jacquemoud and Baret,
1990; Verhoef, 1984; Verhoef, 1985). The aptness of the
methods is analyzed in terms of prediction accuracy for
estimating leaf and canopy chlorophyll content.
Corresponding Author