430 
when most of the remaining maps should be completed. The NMD has re 
cently adopted the concept of a provisional map that will introduce 
certain economies without degrading the map quality, and will allow the 
increased use of automated equipment, which should hold the mapping cost 
per square mile at current levels or less. 
The trends in ADP offer considerable basis for optimism and also a need 
for caution. The capital required today just to get started in digital 
cartography can easily reach several million dollars and there is really 
no way to start on a shoestring. However, once a certain basic threshold 
of capability is reached, there is considerable advantage in adding 
additional capability only when justified and only when advanced capabi 
lity is proven. Maintenance costs on equipment can be significant and 
changes and modifications are frequent. Moreover, everything related to 
ADP hardware appears to be becoming cheaper and more powerful. Even the 
rising cost of developing and maintaining software is leading, as pre 
viously noted, to standardization of modules that can be used effectively 
at relatively low cost. 
The trends most important to a mapping agency seem to be the four we 
discussed—digitizing costs, data base availability, GIS applications, 
and auto-carto applications—and considerable planning is necessary to 
take advantage of technical developments, anticipate future requirements, 
and avoid premature costly commitments. Digitizing costs will probably 
decrease to an acceptable range even for complex data categories, but 
this will require some developments and a few breakthroughs in automated 
and semiautomated techniques. At the moment it appears practical to 
digitize simple categories as fast as possible to build up data base 
coverage and defer the more complex categories for a few years. The NMD 
currently is emphasizing the two DLG categories of boundaries and the 
U.S. Public Land Net and only digitizing hydrography, transportation, 
and other categories when essential for proven requirements. DEM data 
are being gathered with the Gestalt Photomapper as a spinoff to the ortho 
photoquad program, but we expect to digitize the more accurate contours 
from existing maps when the techniques for raster scanning become viable. 
However one cannot afford to be too cautious. There is a need to 
rapidly increase the coverage of data of multiple categories in the data 
base and it is essential to develop automated cartographic applications 
that produce more finished quality maps in a cost-effective manner. 
With some of this strategy in mind, the NMD has recently completed a 
small-scale (1:2,000,000) data base of the entire United States and has 
developed effective techniques to prepare 1:24,000-scale maps using 
automated cartography. The greatest short-term potential gain will be 
in the enhanced ability to respond to requests for custom maps on various 
projections and with a selected treatment of categories. Gradually the 
trends in automated cartography and the decreasing ADP costs will come 
to bear against the map revision issue, especially as data base coverage 
increases. 
The trend in geographic-inforraation-system applications bears careful 
attention since I believe the projected region of cost effectiveness is 
rather late in the decade. There is a temptation to gather level 1 or 
level 2 data now for various display devices and upgrade to level 3 when 
necessary. However, as reported here earlier, the levels do not upgrade 
easily without considerable expense and effort and an initial investment