For the scale 1: 33 000 the opposite result was obtained, possibly 
due to the fact that the pictures were taken in leafing-time and, 
therefore, the leaved crowns appear less distinctly. 
Hilly terrain causes errors unless the differences in level are 
known so that corrections can be applied to the results. Norm- 
ally one has to take level differances into account, especially 
when the whole model is used. The errors increase with increas- 
ing scale. For the Ostad field, which is the most hilly one, an 
error amounting to = 0.5 m has to be calculated with on the 
1:16 000 scale. This has an equalizing effect on the differences 
between scales. 
As is already pointed out, pictures taken with the camera axis 
tilted cause errors in tree measurements. The magnitude of the 
errors varies according to location of the measured tree in the 
stereo model. The errors are most significant around the edges 
of the model. In this investigation the whole stereo model has 
been utilized on the 1: 16 000 scale, while the same ground area 
comprises a much smaller part of the model on the 1:33 000 
scale. This fact may cause a greater mean square error for the 
1: 16 000 scale. An investigation in this respect was carried out 
for the Ostad field, which had the greatest deviations from ideal 
photography conditions. The errors due to tilt have already been 
reported. (Figure 8). By means of these errors, corrections have 
been applied to the measurements in respect of the location of 
the trees in the model. Table 13. 
Table 13 
Systematic and mean square errors before and after corrections for tilt. 
  
  
Before application After application 
of corrections of corrections 
Ope- Scale Single trees Stands Single trees Stands 
rator —— 
Syste- Syste- Syste- Syste- 
Sy Ste M. sq. yste | M. sq. 3 Ste M. sq. yste M. sq. 
matic matic matic matic 
error error error error 
error error error error 
1. |1:16000 —-21 + 1.98 — 16 + 2.81 -18 |+ 201 —10 | +290 
The mean square errors do not seem to be affected by the 
corrections while the systematic error is changed 0.5 resp. 
0.8 m. The reason for this will be that the locations of the trees 
99 
M.