Only roads were used for the positional accuracy 
evaluation of well-defined features. 
Data processing and analysis 
All data were converted to PC Arc/Info coverages. 
Three types of coverage were created: roads for 
positional accuracy evaluation, and lines and 
polygons for attribute accuracy evaluation. 
Attribute accuracy evaluation of linear objects 
After building line topology, the lengths of 
linear attributes contained on the coverages, i.e, 
roads, railway and rivers, were calculated. The 
rate of success was evaluated, and the results are 
presented in table 1. Because of the short time 
interval between the collection of the reference 
and test data (3 years), and the types of object 
considered, we can reasonably assume that no 
attribute changes occurred. 
  
  
  
  
  
  
  
  
  
I.G.N. Map Mono PA-2000 Rate of Success % 
Feature name Distance m. Distance m. Distance m. Mono PA-2000 
Main road 2258.43 2262.40 2261.46 100.17 100.13 
Narrow road 8188.03 9972.27 8289.14 121.79 101.24 
Track & path 21125.36 7589.87 18497.87 35.93 87.56 
Railway 2255.35 225123 2262.78 100.08 100.33 
River 2807.5 2649.07 2691.5 94.36 95.87 
Stream 6479.55 1992.40 2517.86 30.75 38.86 
Total 4311422 2672324 36520.60 61.98 84.47 
  
  
  
  
  
  
  
  
Table 1: Rate of success of line objects 
Attribute accuracy evaluation of polygon objects 
Each land use polygon coverage from both 
monoplotting and stereoplotting was overlaid with 
the corresponding . reference land use coverage 
produced from the IGN map. The polygon areas from 
the overlaid coverage were calculated per area 
attribute. The rate of success was then 
calculated; the results are given in table 2. 
It should be pointed out that for the 
identification and digitization of lines and 
polygons in the digital monoplotting system, no 
magnification was used, while magnification of 4x 
was used on the PA-2000 
  
  
  
  
  
  
  
  
  
LG.N. Map Mono PA-2000 Rate of Success % 
Feature game Area m. Area m. Area m.' Mono PA-2000 
Flood area 256109.8 45054.24. 91373.34 17.59 35.68 
Vineyard 3885623 3612250 3608500 92.96 92.87 
Orchard 573008.8 225816.6 259665.1 39.41 45.32 
Open area 864855.4 45574.76 218962.8 527 25.32 
Forest 470403 338552 352494.3 7197 74.93 
Total 6,050.000 4,267,248 4,530,995 70.54 74.90 
  
  
  
  
  
  
  
  
Table 2: Rate of success of polygon objects 
Analyzing the results in tables 1 and 2, the 
following can be stated. Interpretation using the 
PA-2000 is better than digital monoplotting 
because of stereo viewing and magnification. The 
rate of success with linear features from the 
PA-2000 was 23% higher than monoplotting and close 
to the IGN map (85% success). These very good 
results where obtained, because generally, linear 
features are easier to interpret under high 
magnification and in stereo. 
496 
In table 1, it can be observed that the rate of 
success with "narrow roads" using the monoplotting 
method exceeds 100%, while the "tracks & paths" 
have a success rate of only 36%. This indicates 
that misclassification occurred between "narrov 
roads" and "tracks". The rate of success of 
polygons with PA-2000 was only 4% higher than 
monoplotting. When compared with the IGN map, the 
PA-200 success rate for polygons is only 75%. This 
result is believed to improve if the definition of 
polygon objects is sharpened. 
In interpreting the figures in table 2, we must 
remember that the totals of feature classes are 
influenced by both attribute misclassification and 
attribute change. Some features were obviously 
subject to change, which blurred the attribute 
accuracy of polygon objects. As can be expected 
when using a test site in southern France, the 
vineyards showed the highest rate of success. 
Positional accuracy evaluation 
  
Positional accuracy was evaluated using only road 
features. The line coverage of the digital map 
data was chosen as reference data, since they 
represent the higher-order survey. The digitized 
IGN topographic map was not used for positional 
accuracy evaluation because of possible paper 
distortions. 
The reference coverage in vector format was 
compared with the test coverages collected by 
monoplotting and the PA-2000. Before comparing, 
the test coverages were reformatted from line to 
point coverages. The absolute distances from the 
base coverage to test coverages were calculated. 
The results of the calculations are given in 
figure 2, where the relative frequencies of the 
absolute distances are shown. The statistical 
parameters are given in table 3. 
  
  
  
  
  
  
  
Statistical results (m.) Monoplotting PA-2000 
Maximum distance 39.23 17.72 
Minimum distance 0 0 
Average distance 4.35 1.06 
Standard deviation 3.80 1.09 
Variance 14.41 1.18 
Total digitized points 2310 3.051 
  
  
  
  
  
Table 3: Statistical parameters for positional 
accuracy evaluation. 
The positional accuracy of both methods was 
analyzed using the normal distribution as the 
statistical model. The results of accuracy 
estimation with various confidence levels are 
shown in table 4. The values of the parameter Z in 
this table were drawn from the normal 
distribution. 
Using the 90% confidence level as an example, the 
following interpretation holds for the 
corresponding accuracy values. "The positional 
error at any point is expected to be 9.21 m or 
less for data collected by digital monoplotting, 
or 2.46 m or less for data collected by the 
PA-2000, in 90% of the cases" [1,2].