10. This is in agreement with the 
highest recorded LAI values for winter 
wheat mentioned in literature (Watson 
1971, Langer and Hill 1982). 
2. Negative LAInst values may occur. In 
these cases LAI*»t was set to zero. 
After feeding the measured Vi’s into 
the respective models, a comparison was 
made between the measured LAI's and 
estimated LAI's. 
This was initially done by application 
of the Wilcoxon signed ranks test 
(Siegel and Castellan 1988). This 
distribution-free test is somewhat less 
powerful than Student's T-test for 
paired samples, but requires no 
assumptions regarding normality and 
homogeneity of variances. 
The models yielding significant 
differences between measured LAI and 
LAIont were dropped from further 
analysis. 
As Wilcoxon ' s test does not indicate 
whether the remaining models 
accurately predict LAI, normalized 
averaged deviations (DEVi, 0 »m) of LAIosi 
from measured LAI were calculated as: 
average deviation fron the the 1:1 line 
DEV.. 
average measured LAI 
The normalisation by average measured 
LAI facilitates comparison between data 
from different seasons as the range of 
LAI values was different between 
normally fertilized plots (1984, 1985) 
and the plots receiving a wide range of 
N-fertilizer (1986, 1988). 
Figure 1 summarizes the results of the 
model validation exercise. Following 
conclusions can be drawn: 
1. The SR, ND and TSAVI are valid 
estimators across cultivars and growing 
seasons for pre-senescence LAI of 
winter wheat if no correction for solar 
zenith angle is made. However, the 
accuracy of LAI estimation may vary and 
errors range from 15% to as much as 50% 
of the average LAI. 
2. The SR and TSAVI are valid 
estimators across cultivars and growing 
seasons for pre-senescence LAI of 
winter wheat if the causing variable of 
the model (LAI) is corrected for solar 
zenith angle. The accuracies are 
similar compared with the 
’nocor * 1 -models. 
3. The 'bcor'- and 'vcor'-models are no 
valid estimators of LAI. 
4. SR is the only vegetation index that 
yields valid estimations of 
post-senescence LAI of winter wheat. 
This applies only to the 'nocor'-mode1. 
5. The accuracy of LAI estimation is 
much lower in the post-senescence 
period. 
Remarkably, the LAI of 1986 are not 
more accurately estimated than the LAI 
of the other cultivars, although half 
of the 1986 data served as calibration 
data. 
In general the orthogonal Vi's perform 
poorly. This might be attributed to 
slight architectural differences 
between wheat cultivars for which GRS 
and PVI are more sensitive (Jackson and 
Pinter 1986) . The bad performance of 
PVI apparently corresponds with 
results obtained by Wiegand et al. 
(1979) who found good relationships 
between PVI and LAI, but concluded that 
the slopes differed unexplainably from 
field to field. These differences were 
thought to be caused by complex 
interactions of shadow effects and 
canopy architecture. 
TSAVI, which is labeled as a hybrid 
between a ratio index and an orthogonal 
index, seems to behave as a ratio 
index. 
The proposed RIV and NDIV indices have 
a worse performance than the 
established vegetation indices. 
CONCLUSIONS 
The monomolecular model offers an 
appropriate description of the 
relationship between LAI of winter 
wheat and some vegetation indices. 
The simple ratio (SR) is the only 
vegetation index that can be used 
across cultivars and growing seasons to 
accurately estimate LAI, both in the 
pre-senescence as in the post-senesence 
period. Correcting LAI for solar zenith 
angle or neglecting its effect 
completely in the model yield equal 
results. Other corrections of the Vi's 
for the effect of solar zenith angle 
yield poor results. The ND and the 
TSAVI are equally accurate estimators 
of LAI in the pre-senescence stage. 
Large errors may result from LAI 
estimation of a yellowing wheat canopy. 
It is a fortunate implication for any 
exercise using LAI as an input in yield 
prediction, that the best LAI 
estimation is possible in the 
vegetative stage. 
ACKNOWLEDGEMENTS 
Mr. Peter Haelvoet, Mr. Dirk Tietens 
and Mr. Rik Standaert offered valuable 
assistance for the LAI measurements. 
There was a most pleasant cooperation 
with Mr. Etienne Vanderschueren, owner 
of the farm where the test fields were 
located. 
This research was supported in part by 
IWONL (Comittee for Applied Pedology) 
and NFWO, the Belgian National Fund 
for Scientific Research. 
LITERATURE 
Asrar, G., Fuchs, M., Kanemasu, E.T. 
and Hatfield, J.L., 1984. Estimating 
absorbed photosynthetic radiation and 
leaf area index from spectral 
reflectance in wheat. Agronomy Journal, 
76 (2) : 300-306. 
Asrar, G., Kanemasu, E.T. and Yoshida, 
M., 1985. Estimates of leaf area index 
from spectral reflectance of wheat 
471