For the rows B, C and D the colour is more reddish than for
trees in row A (trees nr.9-12) which has a blue coloured
ploughed field as background instead of red tree canopies.
Damaged trees are masked by these background disturbances.
The case of trees nrs. 7-8 illustrates this point. In camera
position 2 they form a distinct group, but in-position l1 their
specific colour disappears and they show colours which are
more similar to the ones of the healty trees , introducing
a possible omission error in the detection of damaged trees
in camera position 1.
3. Temporal analysis
3.1. Differences in colour and interpretation between temporal
registrations.
It is evident that the colour of the tree canopies on photo-
graphs taken on a different datum will be largely different.
The question however is whether these differences are systema-
tic and whether the colour differences between healthy and
damaged trees remain similar.
The delta diagram of Fig.2 shows the proportion of the densi-
ties in the three emulsion layers for the same trees on diffe-
rent data : august 75 and oktober 77. One of the main troubles
in taking temporal CIR-photographs in our regions is that
good weather conditions are exceptional. Only about 5 to 10
days a year have suitable weather for taking infrared pictures
and there is a low probability of taking photogaphs of exactly
the same phenological conditions each year.
The different symbols of fig.2 refer to trees having a similar
camera position, the letters refer to background colour which
of course, can be very different each year. This diagram shows
that the trees on the photograph taken in 1975 have a more
saturated reddish colour than on the picture of 1977.
The general shift suggests a systematic change of the colour.
However, fig.3 shows clearly that fornoneof the three emul-
sion layers any correlation exists between the densities of