The planimetric coordinates x, y of ihe fixed 
points at the Pecny test field, und thus the coordin- 
ates of the vertices of the 59 imaginary (experimen- 
tal) parcels, had been defined geodetically with a 
mean error of my y = = 2 cm. Hence the acreages P 
of the imaginary 59 parcels, computed according to 
Eq. (1), can be considered correct. 
2P = [Xa(Yn41 —y2-1)] 
2P = [yn(xn-1 —%n+1)] (1) 
The dependence of the areal accuracy of photo- 
grammetrically plotted parcels upon the image scale 
of the aerial photographs and the map scale was de- 
rived from 11 variants of photographs taken at the 
Pecny test fields on the scales of 1:3000 to 1:25,000. 
The test-field points were marked with tempor- 
ary signals of foamed white polystyrene. Those marks 
were centred over the crosses of the granite prisms 
with an accuracy better than 1 cm. 
When checking the areal accuracy of the par- 
cels planimetrically plotted by photogrammetry from 
all photograph variants of the Pecny field, it was 
assumed that 
a) photographs on the scales of 1:3500 and 1:5000 
are used to plot maps on the scale of 1:1000; 
b) photographs on the scales of 1:6800, 1:8800, 
1:9000, and 1:9100, taken with various cameras 
or with cameras having different constants, are 
used for maps on the scale of 1:2000, and 
c) photographs taken with an RC5a Wild camera 
(150/230) on the scale of 1:13,500 and with 
an RC5a Wild camera (110/180) on the scales 
of 1:15,000, 1:20,000, and 1:25,000 are used to 
p'ot maps on 1:5000 scale. 
Plotting, done in 10 different photogrammetric 
laboratories, was carried out on an A7 Wild auto- 
graph in one case, on Zeiss-Jena planigraphs in 3 
cases, and on stereometrographs in 6 cases. Hence 
the results may be considered to give an objective 
picture of the accuracy of photogrammetric practice. 
After numerical plotting of coordinates x(n), 
y(n) and the graphical definition of the coordinates 
x(g), y(g) of the graphically plotted characteristic 
points of the boundary lines, the areas P;(n) were 
calculated using E4. (1) from the numerically de- 
termined coordinates, and the areas P;(g) from the 
graphically determined coordinates of the 76 corner 
points of the 59 imaginary parcels. In every plot- 
ting variant, the deviations of the map areas P;(n) 
end P;(g) from the true parcel area P; were calcul- 
ated: 
APi(n) = Pi(n) —P; 
APi(g) = P,(g) —P;, (2) 
where P; is the true area of the parcel, calculated 
from the geodetical coordinates. Next, the mean devi- 
ations, maximum deviations and mean relative devi- 
ations were calcu'ated for every parcel, according to 
the paragraphs a), b) and c). 
The areal deviations of parcels situated in two 
models could reasonably be expected to be greater 
than those of parcels pictured in a single mode!, par- 
ticularly if calcu'ated from shified coordinates after 
graphical plotting. These two categories of parcels 
were classified accordingly in two groups — A and 
B — for the analysis under discussion. Considering 
the number and magnitude of the deviation in the 
parcels of each of the two groups, weighted arithmet- 
ical averages of aeral deviations AP of the parcels 
were calculated, as well as averages of relative devi- 
ations 6 of the parcel areas on the scales of 1:1000, 
1:2000, and 1:5000. The results are tabulated below. 
  
  
E 
  
  
  
Table 1 
Average deviations of calculated parcel areas after | 
Group photogrammetric plotting 
Mas d ipd CORREA EE n umerical BOMB TI EO graphical : 
i s A 
number AP 5 number | AP | 5 © Qo 
A B of cases 2 0 of cases 2 | o 
[m'] [70] mi |} LA! 
n n | 
1:1000 1 — 14 18 0.42 — | — — 
1:2000 1 o 268 18 | 0.75 111 | 33 1.46 
1:5000 1 - 634 58 1.34 206 | 87 1.99 
| | 
1:1000 2 - - 
1:2000 2 86 38° | 078 66 | 60 1.33 
1:5000 = 2 | 2. Tong 2.32 
| 
  
  
  
  
  
  
  
  
This Table makes it clear that there is no diffe- 
rence in areal accuracy between parcels plotted from 
single models and those plotted from two models, 
provided always the number of cases is sufficient. 
In the further procedure, the areal accuracy of par- 
cels plotted from one and from two models was 
therefore assumed to be practically equal. 
The 59 parcels under investigation were divided 
in 15 groups according to their sizes (cf. Table 2), 
in order to determine the dependence of the areal 
accuracy of photogrammetrically plotted parcels upon 
the image scale and the map scale. In each of those 
groups, weighted averages of deviations from parcel 
areas and weighted aver-ges of the relative devi- 
ations of the areas calculated from photogrammetric 
plottings were computed for each of the image scales 
under consideration.