approximation model of the surface. Photogrammetric DTM's, 
however, suffer from additional instrument and processing 
errors, but are better mathematically modeled and con- 
troiied. The non-stereo approach usins LANDSAT-has mixed 
considerations. 
Another significant problem in the photogrammetric produc-. 
tion of DTM's is the "Bald Earth" problem. The DTM should 
be représentative of the land surface after removal of 
natural landscape and manmade cultural features. In any 
manual or automatic photogrammetric DTM collection system, 
vegetation, such as trees, and buildings, etc., do not al- 
ways allow the elevation of the ground to be measured. 
For example, an elevation profiling scenario in an analytical 
stereoplotter may well pass through an isolated tree, but 
will probably lose its ability to "see' the ground in a 
dense forest area, and the resultant DTM will probably 
have a portion of the tree height combined with the ground 
elevation. In a very dense urban area, a nearlv vertical 
view is required, otherwise the street level may not be 
visible between buildings. The inability to see the "Bald 
Earth" needs to be considered in all DTM accuracy estimates. 
DTM ANALYSIS 
A number of investigators have developed methods for DTM 
accuracy assessment. Although not written for cartographic 
application, Riplev (1981) presents a good collection of 
generic techniques for the analysis of two dimensional 
numerical data, including sampling, smoothing, interpolation, 
and analysis.  Faintich (1983) presents information on inter- 
active analysis of DTM's, and newer results are presented 
in this paper. 
The Defense Mapping Agency (DMA) produces digital data bases 
that describe the physical appearance of the surface of the 
earth. These data bases include, but are not limited to, 
terrain elevation, culture including landscape characteristics, 
end vertical features. This data is collected from digitized 
source maps, from optically or digitally correlated stereo- 
pairs of photographic imagery, and from digital multi- 
Spectral sensor data. A dramatic impact has been made in 
the ability to analyze these digital data bases by applying 
state-of-the-art digital ‘image technology processing and 
display concepts.” These ‘include a variety of color and/or 
black and whire displays of not only intensity/color coded 
matrix data, but also image Processed data using specialize 
convolution filters, texture discrimination. and special 
color representation techniques. "In addition, computer 
generated imagery from these data bases serves as à final 
analysis tool!