1015 
Symposium on Remote Sensing for Resources Development and Environmental Management/Enschede/August 1986 
© 1987Balkema, Rotterdam. ISBN 90 6191 674 7 
The integration of remote sensing and geographic information 
systems 
David G.Goodenough 
Canada Centre for Remote Sensing, Department of Energy, Mines & Resources, Ottawa 
ABSTRACT: The integration of remote sensing and geographic information systems is essential for effective 
resource management. The volume of remote sensing imagery for managing a provincial resource, such as 
forestry, is such that one must use digital image analysis systems. By combining remote sensing image 
analysis and geographic information systems, resource managers can have timely and accurate knowledge of a 
renewable resource. There are, however, several scientific and technical problems that reduce the success of 
this integration. 
This paper describes several integration problems and the LANDSAT Digital Image Analysis System (LDIAS) 
used at the Canada Centre for Remote Sensing (CCRS). Experiments have been conducted integrating a forestry 
geographic information system for the province of British Columbia with LDIAS. Some of the difficulties 
encountered require the use of non-algorithmic solutions which use symbolic reasoning. A brief description of 
expert systems for this integration is given. Several key issues for the future are raised for consideration. 
1 INTRODUCTION 
1.1 Basic Concepts 
A geographic information system (GIS) is a data base 
system for manipulating digital spatial and thematic 
data. Increasingly, such systems are being used in 
developed countries to aid in resource management 
and computer-assisted mapping. A GIS has four 
components (Marble et al 1984): a data input 
subsystem, a data storage and retrieval subsystem, a 
data manipulation and analysis subsystem, and a data 
reporting subsystem. The data inputs are usually 
spatial and thematic data derived from a combination 
of existing maps, aerial photographs, and manual 
interpretations of remotely sensed imagery. With 
the data manipulation and analysis subsystem, the 
user can define spatial procedures to generate 
derived information, such as the best location 
economically to harvest a resource, such as timber. 
The data reporting subsystem is used to generate 
reports in tabular form, digital displays, or maps. 
Because the input data on which a GIS is based 
becomes obsolete quickly, it is essential to update 
periodically the GIS with new spatial and thematic 
data. Remote sensing is often the most cost 
effective source for these updates. 
The information content from large quantities of 
remotely sensed images, such as those derived from 
LANDSAT and SPOT satellites, is best extracted using 
computer systems designed for this purpose. Such a 
system is called an Image Analysis System (IAS). An 
IAS has five elements: data acquisition, 
preprocessing, analysis, accuracy assessment, and 
information distribution. The usual input for a 
remote sensing IAS is a computer compatible tape 
containing a digital image acquired at a remote 
sensing, satellite tracking station. For some 
agencies, corrections for sensor-related radiometric 
and geometric errors will have been performed at the 
tracking station. However, the preprocessing 
subsystem in an IAS usually can perform additional 
radiometric, geometric, and atmospheric corrections. 
These specialized corrections, not normally provided 
by image production systems, include: corrections 
for radiometric distortions due to view angle, 
geometric compensation for terrain relief, 
projection of imagery to a variety of map projec 
tions, and refined atmospheric corrections with the 
aid of meteorological information. The result is to 
give an image as free from errors as possible. This 
preprocessed image is then used for training and 
classification. In a system integrating GIS and IAS 
one could use the GIS thematic data and attributes 
to guide the selection of suitable training areas. 
Often the thematic polygons in a GIS contain several 
spectral and spatial classes. It may be necessary, 
therefore, for the user to have manual control of 
the selection of training areas. After classifi 
cation, the accuracy is assessed using theoretical 
estimation based on class statistics or selected 
test sites derived from ground reference informa 
tion, or both. The derived information can be dis 
tributed in the form of tables, maps, computer 
tapes, or images. The LANDSAT Digital Image 
Analysis System (LDIAS) of CCRS includes all of 
these image analysis functions plus the GIS compo 
nents described above. 
1.2 Integration of Remote Sensing and Geocoded Data 
Figure 1 is a pictorial example of the integration 
of remote sensing with other geocoded data. Stored 
in raster or grid form, one has the LANDSAT and 
aircraft sensor data, digitized aerial photography, 
and topography. Thematic data are represented by 
layers or levels of polygons corresponding to a 
given theme or class. Finally, one has point data 
corresponding to meteorological measurements. This 
figure shows that the integration of GIS and IAS 
must deal with different data representations and 
must provide geometric correspondence or registra 
tion between the different data sets. 
There are two approaches that one might take to 
integrating a forestry GIS and remote sensing. To 
simplify the data representation and labelling, one 
could develop a new GIS having the same raster 
structure as the image data with classes compatible 
with those detected spectrally. Given that one 
LANDSAT Thematic Mapper (TM) image is more than 400 
megabytes, this raster-based unification could lead