649 
model parameters. This resulted into an estimated beet yield of 60.0 tons/ha. Next, SUCROS was calibrated so 
that simulated LAI during the growing season matched estimated LAI values obtained from remote sensing 
measurements as close as possible for all ten fields individually. The WDVI values obtained from the CAESAR 
recordings were used for estimating the actual LAI (section 2.3.1) at three dates during the growing season using 
the fit parameters obtained by Bouman et al. (1992) for sugar beet. The optimization procedure described by 
Bouman (1991) was applied. Results are illustrated in figure 5. On the average, the simulation error of (fresh) 
beet yield decreased from 13.4 tons/ha (17.5%) using ’standard’ SUCROS, to 4.2 tons/ha (5.5 %) with SUCROS 
calibrated to actual LAI values at three dates during the growing season. 
estimated beet yield 
actual beet yield (tons/ha) 
Figure 5. Estimated beet yield using SUCROS calibrated to measured LAI versus actually obtained beet yields. 
5 - SUMMARY AND DISCUSSION 
A framework was presented to integrate crop canopy information derived from optical remote sensing with crop 
growth models for the purpose of growth monitoring and yield estimation. Basic for this framework is that (bi-) 
directional optical measurements by performing observations at two viewing angles are combined with high spectral 
resolution observations performed with an imaging spectrometer. The former are used for estimating LAI and 
LAD simultaneously; the latter are used for estimating the leaf scattering coefficient. These are the crop parameters 
that play an important role in both the processes of crop growth and canopy reflectance. Subsequently, the estimated 
crop parameter values were used as input into crop growth models and for calibrating crop growth models. 
The framework was applied to data gathered during the MAC Europe 1991 campaign over the Dutch test 
site Flevoland. Results for sugar beet indicated the feasibility of estimating LAI, LAD and leaf optical properties 
from reflectance measurements. A critical point to consider is the precision and additional value of the parameter 
values derived from remote sensing compared to the standard values already used in the growth model. For instance, 
much relative benefit might be obtained from the estimation of leaf colour expressed in leaf nitrogen or chlorophyll 
content. Especially the modelling of leaf nitrogen status in canopies is extremely complicated (but equally important 
through its effect on maximum leaf photosynthesis rate) and actual information derived from optical reflectance 
would be valuable. 
The method of model calibration was tested on sugar beet. The simulated yield was clearly better in agreement 
with actually obtained yields after model calibration than without model calibration. Since the calibration procedure 
mainly concerned the calibration of the simulated LAI, these results indicate the importance of LAI for accurate 
growth simulation. 6 
6 - ACKNOWLEDGEMENTS 
W. Verhoef is acknowledged for providing the SAIL model and B. Bouman and J. Goudriaan are acknowledged 
for providing the SUCROS model. I am very grateful to S. Jacquemoud and F. Baret (INRA, Montfavet - France) 
for providing the PROSPECT model. NASA and ESA are acknowledged for providing the AVIRIS data in the 
framework of MAC Europe 1991. This paper describes a study that was carried out in the framework of the NRSP-2 
under responsibility of the Netherlands Remote Sensing Board (BCRS) and under contract no. 4530-91-11 ED 
ISP NL of the Joint Research Centre.