67 
.cal Research, Vol. 
Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986 
cloud 
ices. Submission to 
lospheric effects 
the thermal 
on satellites. The 
iadiometer (AVHRR), 
No. 1, p. 161- 
Base map production from geocoded imagery 
Dennis Ross Rose & Ian Laverty 
Macdonald Dettwiler and Associates Limited, Richmond, B.C., Canada 
Mark Sondheim 
British Columbia Ministry of Environment, Surveys and Resource Mapping Branch, Victoria, B.C., Canada 
ABSTRACT: Current techniques of producing topographic base maps rely chiefly 
on aerial photography as the data source. Topography and planimetry are 
derived using standard photogrammetric methods and manually operated equipment. 
Although this method meets current accuracy requirements when used properly, it 
has several inherent problems. For instance, the collection of source data is 
costly as it involves special aircraft collection missions. Extraction of 
relief and feature detail using manual methods is prone to operator error. 
Solutions to these problems must provide more cost-effective means of obtaining 
data and a more automated approach to information extraction from this data. 
During the past decade much progress has been made in the correction and 
processing of digital satellite imagery. Techniques such as geocoding have 
been developed and refined to the stage where map accurate imagery can be 
produced in a high-throughput environment. It is possible for mapping agencies 
to use this imagery as an input to many standard mapping operations. Using 
technology which is available today, it is possible to accurately derive 
planimetric, thematic, and topographic maps using this digital imagery as the 
primary source of information. 
This paper will discuss the recent advances in technology which have provided 
the means to produce such accurate imagery as well as describe how this imagery 
can be used to produce quality map products in a production environment. 
Topics such as feature extraction, thematic classification, elevation 
derivation, and resultant map accuracies will be discussed. A sample 1:50,000 
scale base map, derived entirely from LANDSAT TM imagery, will be presented and 
evaluated to demonstrate the concepts discussed. 
BACKGROUND 
The year 1986 will prove to be a very 
significant one to both the remote sensing 
and cartographic communities. The recent 
launch of the SPOT satellite and the ability 
to generate LANDSAT-TM and SPOT precision 
geocoded products will, for the first time, 
allow cartographers to use 
digitally-processed satellite imagery as a 
prime data source for map production. 
Satellite imagery is economical and provides 
high information content. Production 
oriented mapping systems are now being 
developed which extract feature and relief 
content entirely from imagery. General 
acceptance of these systems will only come 
about once it can be shown that the 
satellite derived map products compare 
favorably to those generated by typical 
photogrammetric techniques. 
Current mapping procedures are based on 
photogrammetric methods and meet required 
accuracy standards. However, several 
problems exist with such methods. The 
procedure is slow and data acquisition is 
expensive. Resultant quality may vary from 
interpreter to interpreter. 
Developed nations often have well developed 
mapping programs. Such countries continue 
mapping to update existing maps or to 
improve the accuracy of existing maps. 
Developing nations frequently have not been 
mapped completely, and any existing maps may 
not be of acceptable quality. Such 
countries cannot afford the time (often 
decades) or cost required to produce a 
quality map data base. 
Digital satellite imagery should provide 
help for both of these cases. Satellite 
imagery provides very wide region coverage. 
A single LANDSAT-TM image covers an area of 
34,000 sq. kilometres, compared to a 130 
sq. kilometre aerial photograph (1:50,000 
photo scale). Similarly, the amount of 
ground control required to correct a 
satellite image is one tenth of that 
required to correct a corresponding area of 
aerial photographs. These differences imply 
a highly reduced cost of data acquisition 
when comparing satellite imagery and control 
to aerial photography and control. 
The digital and multispectral nature of 
satellite imagery also provides advantages 
over aerial photographs. Using digital 
techniques it is possible to ~ extract' 
terrain, planimetry, and landcover 
information in an automated fashion from the 
imagery. Since the infomation is digital, 
it can be used easily with other digital 
data (e.g. digital maps). Finally, 
automated map production will lead to more 
consistant quality map production, as human 
intervention is reduced.