12) may be 
Therefore, 
every point 
Iculated by 
xt round. 
ise factors, 
or them by 
able condi- 
resent fine 
lines sheet 
cal plotter 
2km wide, 
y mistakes 
already re- 
ts that ob- 
e 9, one in 
80 contour 
out 38,000 
n the fol- 
and many 
npare the 
factors is 
2). 
Table 2 Case factors 
  
  
  
  
  
  
  
  
factors standard value | case| changed value 
(a) 2nd order 1 3rd order 
2 Ist order 
(b) 50m 3 25m 
4 100m 
(c) 50m 5 100m 
(d) a ,=5m, 6 ,=2m 6 c -lm,c,-1m 
7 c -5m,c,-im 
| ©) concerned 8 not concerned 
(f) 10m 9 |5m 
(g) 3 10. 1.5 
14 |1 
  
  
  
  
  
  
(a) Observation equations in local DTM generation 
The order of the polynomial used for observation equa- 
tions is changed in the case 1 and 2. The higher order polyno- 
mial is used, the more flexible DTM is generated, however, the 
data are not corrected very much. 
(b) Weights of observations in local DTM generation 
Weights of observations are calculated by (2.6). The pa- 
rameter d, is changed in the case 3 and 4. The larger standard 
distance is taken, the more observations affect the movement on 
each point, and the gentler drawing will be made as the result. 
(c) Area for collection of observations 
The radius of the area for correction of observations (r in 
Fig 2) in local DTM generation is changed in the case 5. 
(d) Observation accuracy in planimetry and elevation 
Observation accuracy is assumed from observing situation 
(9,6, 11(2.12)). The less accurate observation in planimetry 
in comparison with one in elevation is assumed, the larger 
movement will be given to each point. 
(e) Concern about the accuracy of generated DTM 
In the case 8, the accuracy of local DTM( c ) is not con- 
cerned in the step of correction (2.10). 
(f) Point pitch of contour lines 
The point pitch affects the number of observations in DTM 
calculation. Besides, the higher density data present the 
smoother drawing as the result, however, calculation will take 
longer. 
(g) Iteration 
Number of iteration is changed in the case 10 and 11. 
865 
3.3 Results 
The edited drawing by each case is shown in Fig 6. Calcu- 
lation time was about 15 minutes for case 9, and 8 minutes for 
each of other cases. [computer: Sun sparc station 2 (28 mips), 
CPU time] 
4. CONCLUSION OF LOCAL DTM EDITING 
Local DTM editing by the standard condition presents 
properly smoothed drawing as the result, which would be ac- 
ceptable as the base map of 1:25,000 topographic map. The 
expression of the terrain by manual editing would not be far 
different from one of this method. Considering about its effect 
for reducing of editing work, this Local DTM method is valu- 
able. Actually editing by this method has already put into prac- 
tice at the Geographical Survey Institute of Japan. 
In addition, following facts appeared over these examina- 
tions. 
(a) 1st order DTM editing does not fit on some parts of 
terrain such as where a deep valley or a steep peak exists, on the 
other hand 3rd order DTM editing still have some unsmoothed 
parts left. 
(b) Assumed accuracy of observations is especially affec- 
tive for the results. In the case 6 in which standard deviations of 
both planimetry and elevation observations are assumed as very 
small(1m), edited drawing has only a little change, on the con- 
trary in the case 7 in which the s.d. of elevation observation is 
assumed to be 5 times large as one of planimetry observation, 
edited drawing has greatly moved to be very much loose. 
(c) Area size for collecting observations at local DTM 
generation does not cause much difference as far as it is between 
25m and 100m in radius. 
(d) The drawing becomes looser every after one cycle of 
iteration, however, there is little change over 3rd iteration. 
The results mostly satisfy our demands, however, there are 
some tendencies and problems to be concerned by followers. 
(a) By local DTM editing, ridges in the drawing have been 
a little widened and valleys have been narrowed. 
(b) Position of mountain peaks is not fixed during calcula- 
tion, therefore edited drawing may present those peaks shifted a 
bit. 
(c) In observation, neighboring points on the same contour 
line are correlated to each other, however, each of them is con- 
cerned independently at the correction process. That may make