Clevers, Jan 
  
IE 
This paper will focus on two aspects of the project: 
e Application of a semi-empirical approach for estimating the leaf area index (LAI) from high resolution optical 
satellite data at different moments during the growing season. Results are compared with field measurements. 
e Assimilation of the LAI estimates into a calibrated crop growth model. The simulated LAI and final yield are 
compared with field measurements. 
1.2 Crop Yield Prediction 
Monitoring the growth of agricultural crops during the whole growing season is important for observing anomalies in 
growth and development in order to be able to adjust the growth of the crops, and to obtain yield predictions well before 
harvest time. More and more use is being made of crop growth models for this purpose. Remote sensing data can be 
used for calibrating growth models for actual field conditions, thus rendering the combination of growth models and 
remote sensing data a valuable tool for growth monitoring (Maas, 1988; Bouman, 1991; Delécolle et al., 1992; Clevers 
et al., 1994). 
A key variable used for calibrating a crop growth model is the LAI. The LAI may be obtained from optical remote 
sensing data by analytically inverting a radiative transfer model. However, best results are obtained by using a semi- 
empirical approach (using a vegetation index) in estimating the LAI regularly during the growing season (Clevers et al., 
es in 1994). 
well 
mote Mechanistic and dynamic crop growth models describe the relationship between plant physiological processes and 
x the environmental factors such as solar irradiation, temperature, and water and nutrient availability. They simulate daily 
3 the growth and development of crops, and finally a simulation of yield at harvest time is obtained. As such these models 
1. In may also be used for yield prediction early in the growing season by assuming average meteorological conditions. 
area However, it is advisable to “check” modelling results with some sort of information on the actual status of the crop 
i on throughout the growing season. Optical remote sensing can provide such information (Clevers, 1988; Bouman, 1991). 
1996 One possibility is to force the model simulation through obtained LAI estimates by resetting the simulated LAI value to 
ages the retrieved LAI value at a remote sening event. Another possibility is to force interpolated LAI values retrieved by 
well remote sensing. 
field 
2 METHODOLOGY 
First, a simple reflectance model (CLAIR) is used for estimating the LAI from optical data. Subsequently, these LAI 
values are forced on the calibrated simulation model in two ways: 1) A reset of the simulated LAI values by the LAI 
values retrieved from three of the four SPOT images (3x reset), and 2) force the interpolated SPOT LAI values 
continuously on the simulation model (every day, between 1“, 2*4 3*3 and 4^ SPOT image). 
ion, 
ated 2.1 Estimating Leaf Area Index 
n 
Lu Complex physical reflectance models usually simulate reflectance for varying crop characteristics and incorporate 
ing simplified structural and optical properties of the canopy. Such models are very useful for sensitivity analyses in order 
to investigate the influence of crop characteristics and external variables on reflectances or simple vegetation indices. 
Moreover, they are very valuable for investigating boundary conditions to the application of certain simplifying 
relationships. However, they are difficult to invert and too complicated for practical applications. 
Km Clevers (1988, 1989) derived a simplified, semi-empirical reflectance model for estimating LAI (CLAIR model). In this 
ts model, first, the WDVI (weighted difference vegetation index) is ascertained as a weighted difference between 
xels measured near-infrared (NIR) and red reflectances, assuming that the ratio of NIR and red reflectances of bare soil is 
constant. In this way a correction for the influence of soil background is performed: 
sing 
sing WDVI = NIR - (C x Red) (1) 
ical 
NIR = total measured NIR reflectance; 
eric Red = total measured red reflectance; 
ture C = slope of the (soil-specific) soil line, or ratio between NIR and red reflectance of soil. 
in Subsequently, this WDVI is used for estimating LAI according to the inverse of an exponential function: 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B7. Amsterdam 2000. 273