I 
en vegeta- 
rsus diffe 
MILKY 
MATURITY 
RHYS IOl0 
GICAl HA 
TURITY " 
v 
I 
250 
ated vege- 
vegetation 
g to the 
;es. From 
lase, the 
on indices 
i harvest, 
all the 
the same 
point out 
vegetation 
, if the 
iloited to 
e sensing 
observing 
L profiles 
tion cover 
;ages. As 
ion (R) is 
vegetation 
functions 
vident in 
3ver. That 
ig phase. 
.ated with 
: for soil 
leading is 
Leld (if a 
lition is 
assumed until the harvest) it is essential to adopt 
indices yielding the maximum variance between the 
index values of the wheat fields. In Fig. 2 this 
fact is illustrated taking into account three 
spectral vegetation indices; the maximum percentage 
variance at heading between index values of the 
wheat yields is related to the ratio index (R). The 
minimum variance between the field's index values, 
corresponds to the maximum correlation with the 
final grain yield. 
Considering the ideal spectral profile plotted in 
Fig. 3 from tillering to physiological maturity, it 
is possible to identify three spectral profile 
regions where the vegetation index families work to 
the maximum efficiency. 
From tillering to the end of the elongation stage, 
the biophysical parameters closer to the final^grain 
yield of the wheat, is the n° of plants per m . In 
addition, during this time interval, % of soil 
vegetation cover ranges from 10% to 45%. Thus the 
reflectance of the soil is substantial portion of 
the total radiance captured by the radiometre. The 
perpendicular index yield values are less influenced 
by the soil component. 
From booting to heading, the soil vegetation cover 
increases to 85%, the spectral component of the soil 
is not as influential as in the previous stages and 
the perpendicular indices are not as efficient as in 
the ratio family. During these stages the 
biophysical parameter most correlated to the final 
grain yield can be considered the leaf area index, 
that is closely tied to the soil vegetation cover. 
It is not easy to explain why the ratio family is 
more efficient than the vegetation indices belonging 
to the other ones. One explanation is that the ratio 
indices are less influenced by collateral effects 
due to the plant architetture. 
From flowering to the soft-dough stage the leaf 
yellow component increases. In this time interval it 
seems that the vegetation indices based on the 
difference concept yield the maximum correlation 
between the index values and the final grain yield 
of the wheat field. No satisfactory explanation 
exists in this case either, although it is possible 
to relate the grain-filling-process to the duration 
of the green leaf area index. This parameter can be 
detected more efficiently by using indices based on 
the difference concept rather than on the other 
index families. 
At the physiological maturity all the plant leaves 
are yellowed and collapsed so that all the 
vegetation indices proposed in this study have ^a 
significant correlation with the n° of plant per m . 
4 CONCLUSIONS 
Wheat yield forecasting directly supported by means 
of vegetation indices is still in a preliminary 
phase since a significant correlation between 
spectral vegetation index values and final grain 
yield exists only at the heading phase. Moreover 
this study has pointed out that the most appropriate 
vegetation index can be selected only by considering 
the phenological stage of the wheat crop. Although 
it is not indicated in this report, the integrated 
values of the vegetation indices have been 
calculated and related to the final grain yield. The 
use of the vegetation index duration values does not 
significantly improved the relationship between the 
remote sensing data and the biophysical parameters 
of the wheat crop. On the other hand due to 
cloudness in the area the quantity of space remote 
sensing data is limited. Thus the possibility of 
carrying out multitemporal observations in Central 
and Northern Italy is rather low. 
5 REFERENCES 
Aase, J.K., and Siddoway, F.H. (1981): Assessing 
winter wheat Dry-Matter production via Spectral 
Reflectance Measurements. Remote Sens. Environ. 
(11): 267-277. 
Badhwar, G.D., and Shen, S.S. (1984): Techniques for 
the estimation of leaf area index using spectral 
data, Proc. Symp. Machine Proc. Remote Sensing 
Data, Pordue Univ., West Lafayette, Indiana: 333- 
338. 
Daughtry, C.S.T., Gallo, K.P., Bichl, L.L., Kanema- 
su, E.T., and Asrar, G. (1984): Spectral estimates 
of agronomic characteristic of crops, Proc. Symp. 
Machine Proc. Remote Sensing Data, Pordue Univ., 
West Lafayette, Indiana: 348-355. 
Goel, N.S., Henderson, K.E., and Pitts, D.E. (1984): 
Estimation of leaf area index from bidirectional 
spectral reflectance data by investing a canopy 
reflectance model, Proc. Symp. Machine Proc. Remo 
te Sensing Data, Pordue Univ., West Lafayette, In 
diana: 339-347. 
Tucker, C.J., Holben, B.N., Elgin, J.H., and McMur- 
trey III, J.E. (1980): Relationships of spectral 
data to grain-yield-variation, Photogram. Eng. and 
Remote Sensing (46): 657-666. 
Tucker, C.J., Holben, B.N., Elgin, J.H., and McMur- 
trey III, J.E. (1981): Remote Sensing of total 
Dry-Matter accumulation in winter wheat, Remote 
Sens. Environ. (11): 171-189. 
267