228 
sured. The subjective and qualitative nature of the 
first approach and the destructive sampling of a 
few plants, resulting in a large variability be 
tween measurements, of the second approach are the 
main disadvantages. 
Remote sensing techniques provide quantitative in 
formation about a whole field trial instantaneously 
and non-destructively. The aim of the research of 
this paper was to establish whether remote sensing 
can support and/or replace conventional field mea 
surements in field trials. Emphasis was on impro 
ving precision in field trials. 
2 METHODOLOGY 
2.1 Aerial photography 
Black and white multispectral aerial photography 
was carried out by taking vertical photographs from 
a single-engine aircraft, equipped with Hasselblad 
cameras with Planar 100 mm lenses, at intervals of 
about two weeks. A Minolta photometer was used for 
correcting the camera settings for differences in 
illumination between missions, and the expected 
reflectance level of the objects was also taken 
into account for obtaining correct levels of ex 
posure. 
Missions were only carried out under nearly con 
stant atmospheric conditions, and during a mission 
camera settings (exposure time and relative aper 
ture) were not altered for a certain film/filter 
combination. By a judicious choice of aerial films 
with satisfactory sensitivity used in combination 
with filters, the optimal wavelengths identified by 
Bunnik (1978) were matched: a green band (555-580 nm), 
a red band (665-700 nm) and an infrared band (840- 
900 nm) (see Clevers, 1986a). The development of the 
films was standardized. 
Densities were measured by means of an automa 
tized Macbeth TD-504 densitometer with an aperture 
of 0.25 mm. After converting the densities to radi 
ant energy values by means of the characteristic 
curve, compensation for off-axis imagery was made 
for each image point (only for light fall-off; vig 
netting was avoided). There exists a linear rela 
tionship between this corrected radiant energy value 
and the reflectance of the object (Clevers, 1986a, 
1986b). In this linear function the exposure time, 
relative aperture, transmittance of the optical 
system, irradiance, path radiance and atmospheric 
attenuation were incorporated. Reference targets 
with known reflectance characteristics were set up 
in the field during missions and recorded at the 
same camera setting and under the same atmospheric 
conditions as the field trials, in order to ascer 
tain the parameters of the linear function. 
Reflectance factors ascertained in this way with 
multispectral aerial photography appear to be well 
calibrated when compared with MSS and radiometers 
in the field (see Clevers, 1986b), supplying infor 
mation that is only influenced by the object. This 
allows multitemporal comparison and analysis. 
2.2 Test area 
The research was carried out at the ir. A.P. Minder- 
houdhoeve, experimental farm of the Wageningen Agri 
cultural University (The Netherlands), situated in 
one of the new polders, Oost-Flevoland, which was 
reclaimed about 30 years ago. The new polders are 
flat, uniform and highly productive agricultural 
lands with a loamy topsoil. 
2.3 Experiments 
The usefulness of remote sensing techniques in agri 
cultural research was verified by means of analysing 
several field trials. 
Field trial 116 in 1982: Barley in a split-plot de 
sign in three replicates (cultivar "Trumpf"). Whole- 
plot treatments were 2 sowing dates: 26 March (Zl) 
and 22 April (Z2). Split-plot treatments were 6 ran 
domized nitrogen levels: 0-20-40-60-80-100 kg N per 
ha (N1 to N6). 
Field trial 116 in 1983: Barley in a split-plot de 
sign in four replicates (cultivar "Trumpf"). Whole- 
plot treatments were 2 sowing dates: 7 March (Zl) 
and 21 April (Z2). Split-plot treatments were 5 ran 
domized nitrogen levels, applied as split dressing 
(in kg N per ha): 
before sowing 
Feekes stage 7 
Total 
N1 
0 
0 
0 
N2 
20 
0 
20 
N3 
20 
20 
40 
N4 
20 
40 
60 
N5 
20 
60 
80 
For Feekes stages see Large (1954). 
Field trial 92 in 1982: Spring wheat in a split-plot 
design in three replicates (cultivar "Bastion", so 
wing date: 26 March 1982). Whole-plot treatments 
were 2 fungicide treatments: no fungicides at all 
(F0) and 4 kg Bavistin M per ha at Feekes stage 
5 combined with 0,5 1 Bayleton per ha at stage 10.4 
(FI). .Split-plot treatments were 2 sowing densities 
and 4 nitrogen levels, which were completely random 
ized within the whole plots. The sowing densities 
were: 150 seeds per m * 2 3 (SI) and 300 seeds per m 2 
(S2). Only results for one nitrogen level will be 
presented (20 kg N per ha before sowing combined 
with 80 kg N per ha at F8). 
Field trial 92 in 1983: Winter wheat in a split-plot 
design in three replicates (cultivar "Arminda", so 
wing date: 25 October 1982). Whole-plot treatments 
were 2 fungicide treatments: no fungicides at all 
(F0) and 3 kg Bavistin M per ha at F5 combined with 
2 kg Bayleton CF per ha at F10.4 (FI). Split-plot 
treatments were 3 sowing densities and 4 nitrogen 
levels, which were completely randomized within the 
whole plots. The sowing densities were: 150 seeds 
per m 2 (SI), 300 seeds per m 2 (S2) and 600 seeds 
per m 2 (S3). Only results for one nitrogen level 
will be presented (120 kg N per ha at F8). 
Field trial 95 in 1983: This trial was similar to 
field trial 92 in 1983. The cultivar of winter wheat 
was "Okapi" and the sowing densities were: 100 seeds 
per m 2 (SI), 200 seeds per m 2 (S2) and 400 seeds per 
m 2 (S3). Only results for one nitrogen level will be 
presented (40 kg N per ha before sowing combined 
with 100 kg N per ha at F8). 
3 RESULTS 
3.1 Spectral Reflectances 
In field trial 116 in 1982 reflectance factors were 
ascertained by means of multispectral aerial photo 
graphy in green, red and infrared spectral bands. 
The effects for two nitrogen levels for two sowing 
dates are illustrated in figure 2 for all three 
bands. In this figure the reflectance is plotted 
as a function of days after sowing. The patterns 
of green and red reflectances were similar. The re 
flectance in the visible bands decreased with in 
creasing growth during the beginning of the growing 
season. On the first recording date (28 April) only 
bare soil was present, resulting in similar reflec 
tance values in all spectral bands considered. At 
complete soil cover the reflectance in the visible 
bands remained fairly constant. At the end of the 
growing season this reflectance increased due to 
senescence of the crop. The sowing date effect was 
evident. The late-sown crop showed a faster develop 
ment due to higher temperatures during germination 
and initial growth than the early-sown crop, re 
sulting in a shorter growing season (shift in re 
sponse). There were significant (negative) nitrogen 
effects on reflectances in the visible bands, but 
the differences were small. For the infrared reflec- 
. a 
GREEN RE 
b 
RED REFL 
20 
15 
10 
0 
0 
c 
INFRARED 
60 
50 
40 
30 
20 
10 
Z2 
0 
0 
Figure 2 
(c) infi 
and Z2) 
Field tr 
tance th 
was evic 
red spe( 
visible 
trogen i 
throughc 
In fie 
two nitr 
trated j 
sowing c 
was a ne 
(mostly 
season t 
frared r 
dates. I 
less prc 
In adc 
soil mo] 
son in 
moisture 
all thre 
tively j 
top soil 
the earl 
(for the 
tained c 
after s< 
spective