Range (m) eo6 eo7 eo8 eo9 | eo10 
  
  
  
  
- 10.05 16.9 16.4 16.7 15.8 |. 17.1 
+0.10 31.9 31.1 32.2... 31.4 | 32.1 
ed 10.15 45.5 43.7| 46.2| 47.0| 46.6 
10.20 56.7 5701 58.7| 59.0 | 59.0 
  
Under -0.20 11.6 11.5 14.21 11:017 41:2 
Over 40.20 31.7 | 31.5 | 296 | 300 | 298 
  
  
  
  
  
  
  
  
  
Table 6 ERDAS DEM Comparison Statistics, 
Beach Only (96 of beach points) 
  
Range (m) | is01 | is02 | is03 | is04 | is05 | is06 
  
  
  
  
10.05 84| 75 74! 93] 87! 104 
+0.10 15.5 | 16.3 | 16.8 | 19.3 | 18.4 | 19.9 
+0.15 23.4 | 24.2 | 26.2 | 27.8 | 28.7 | 30.1 
+0.20 31.6 | 32.0 | 32.6 } 37.6 | 38.0 | 38.3 
  
Under -0.20 | 33.5 | 29.4 | 27.8 | 23.0 | 21.8 | 20.6 
Over +0.20 | 34.9 | 38.6 | 39.6 | 39.4 | 40.2 | 41.0 
  
  
  
  
  
  
  
  
  
  
Table 7 ImageStation DEM Comparison Statistics, 
Cliff Only (% of cliff points) 
  
  
  
J ANR 
  
  
  
  
  
  
  
  
  
  
  
  
eo1 eo2 eo3 eo4 eo5 
+0.05 9.1 9.3 8.3 10.7 8.5 
+0.10 17.8 17.0 14.9 |. 20.4 |- 17.8 
10.15 27.8 24.7 22.9 28.1 | 26.9 
+0.20 35.9 34.9 29.6 37.1 | 35.7 
Under -0.20 21.6 21.5 28.7] 20.6] 212 
Over «0.20 42.5 43.6 41.7 | 42.3 | 43.1 
  
  
  
  
  
  
y Table 8 OrthoMAX DEM Comparison Statistics, 
M Cliff Only (96 of cliff points) 
ce 
iff Range (m) eo6 eo7 eo8 eo9 | eo10 
1e X 0.05 9.3 9.7 1.7 8.5 8.0 
34 10.10 18.3 18.1 14.8 17.9: 16.3 
+0.15 26.9 | 266 206 | 26.4 | 224 
+0.20 36.0 86.31 28.6 3611| 31.1 
  
Under -0.20 21.8 20.6 31.5] 22.3] 28.6 
Over 40.20 42.2 43.1 39.9| 41.6| 40.3 
  
  
  
  
  
  
  
  
  
Table 9 ERDAS DEM Comparison Statistics, 
Cliff Only (96 of cliff points) 
  
The optimum DEMs for the beach surface are is01, is02, 
eo3 and eo10. Both is01 and is02 have been generated 
using a flat terrain setting with a high and medium 
smoothing factor, respectively. Not surprisingly, the 
poorest ImageStation DEMs have been generated from 
the Hilly and Mountainous settings. However, it can be 
seen that is04 has similar statistics to both is01 and is02, 
and has been generated from a Hilly setting, medium 
Smooth parameter. The optimum ERDAS DEMs have 
been generated using the larger template sizes and lower 
parallax values (see Table 2 and 3). 
) 
1 
6 
0 
7 
1 
9 
The optimum DEMs for the cliff face are is06, eo4 with 
slightly poorer results from is04, is05, eo6 and eo7. The 
particular ImageStation DEMs have been generated from 
both Hilly and Mountainous settings; the ERDAS DEMs 
from smaller minimum and maximum template sizes. 
923 
However, it is interesting to note that an increase in the 
maximum template size, eo4, and not the maximum 
parallax, eo6, improved the overall correlation. 
  
    
   
Northings | 
1 Bottom of Cliff — gp 
420500m 7] Top of Cliff Peg 
  
  
TT Er TTT rT TT TT Tr TTT TTT] 
539400m 539600m 539800m 534000m 
Eastings 
  
  
  
Figure 1 Height Difference Distribution Map (is01) 
The Tables have shown the percentages of points within 
various ranges but it is also important to identify the 
distributions of the height differences. Figure 1 is a 
typical example of the distribution of height differences 
where the approximate top and bottom of the cliff line 
have been identified. A full analysis of these distribution 
plots is still in progress to identify the correlation between 
the magnitude of the height difference and the 
topography and image characteristics. It is interesting to 
note from just this single example that there are areas 
with greater than x0.2m. in all three coastal zones 
(beach, cliff and cliff top). 
5. CONCLUSIONS 
The results from these simple investigations have 
revealed some interesting general features of the 
automated digital elevation modelling process. The 
choice of parameter settings is important to achieve the 
optimum results. Choosing the appropriate simple 
terrain defining parameters in the ImageStation does 
consistently improve the quality of the results. The 
ERDAS system appears to be less predictable when 
changing the variable parameters. However, the analysis 
is still being undertaken. 
The research project is to continue in greater detail to 
establish some criteria for using these DEM processes in 
the coastal zone. This analysis must be matched with 
the practical requirements of the environmental scientists 
and may result in a compromise in terms of a rapid data 
capture and processing technique, and the quality of 
result obtainable. 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996