ANALYTICAL TOOLS AND PROCEDURES FOR GEOINFORMATION PROCESSING 
J.A.R. BLAIS 
Department of Surveying Engineering 
The University of Calgary 
Calgary, Alberta, Canada 
ABSTRACT 
With the rapidly evolving possibilities in computations and visualization as a consequence of tremendous changes in computer 
technology, the processing of geoinformation can be increasingly automated with adaptive procedures in varied applications. 
Selecting appropriate methodologies usually involves dealing with incomplete information while minimizing the necessary 
assumptions. One general approach which has proven successful in very different applications is using information theory. Practical 
applications of information theory in spectrum estimation, adaptive filter design and inverse problems will illustrate the potential for 
other similar applications. General conclusions and recommendations will be included. 
1. INTRODUCTION 
Fundamental changes are taking place in positioning, mapping 
and related fields. Angle and distance measurements are being 
supplemented and replaced by the recording of observations 
from satellites, accelerometer systems and gyroscopes along 
with frequency standards for digital signal processing. Analog 
photographs are rapidly becoming obsolete with digital imaging 
systems and mapping operations integrated into geographical 
and more generally spatial information systems. Visualization 
systems will soon replace conventional topographical maps with 
terrain rendering and spatial information display systems. These 
fundamental changes have far reaching implications for the 
analytical tools and procedures which are necessary in the data 
processing of observations and measurements. 
The advent of modern computer technology with the ever 
increasing computing power and availability also has profound 
implications in the selection of tools and procedures. In 
particular, algorithms are becoming more sophisticated and 
adaptive procedures are most often desired in view of 
automating the processing of observations and measurements. 
The increasing sophistication has led to dealing with patterns, 
trends and the like in order to have adaptive procedures. These 
tendencies are really leading to information processing in 
general and even knowledge processing in specialized 
application areas. 
Among the outstanding questions in this field of geoinformation 
processing are the identification and recognition of patterns, the 
quantification of information contents for adaptive processing 
and quality control, and inference procedures that can transcend 
the data processing requirements. In other words, what are the 
new requirements in terms of tools and procedures for 
geoinformation processing that will lead to the knowledge- 
based systems of the near future? 
The following discussion will first consider the mathematical 
differences between data, information and knowledge, and then 
the relationships between accuracy, uncertainty and semantics. 
These concepts have important implications for the 
understanding of what is needed in categorizing and 
characterizing the analytical tools and procedures for 
geoinformation processing. As distinct from data processing, 
information processing deals with adaptive methods of filtering, 
pattern extraction and processing, dealing with incomplete 
information for decision support and other applications. 
A brief overview of information theory which has proven very 
appropriate for quantifying information contents and 
categorizing patterns and structures will be given. Examples 
from three areas of applications will be discussed to illustrate 
the use of information theoretical principles in those contexts. 
General conclusions and recommendations will be included as 
guidelines for other areas of potential applications. 
2. FROM DATA TO INFORMATION 
Observations and measurements are well known to provide little 
information unless they are properly designed and adequately 
carried out. This clearly illustrates that data do not always imply 
185 
substantial or significant information and hence proper 
strategies and analyses are required. Among the principal 
objectives of geoinformation processing, the extraction of 
useful information from available data ranks very high and 
much research and development efforts have been invested in 
various application fields. However, fundamental questions 
remain about the nature of information and its quantification or 
measure to decide on optimal data and information processing 
strategies for all kinds of applications. 
In the case of distance and angle data processing for positioning 
purposes, the analysis is definitely simpler than when dealing 
with pattern identification and recognition in digital image 
processing. Accuracy and reliability in survey networks are well 
understood but the analogous concepts with digital imagery and 
spatial information are definitely more complex. With the 
current emphasis on spatial information systems (SIS), these 
questions are becoming more and more important and further 
investigations are definitely warranted. 
Data are observations or measurements that are collected in 
order to extract some required or expected information and 
knowledge. Data processing using conventional algorithmic 
methods is well understood in physical science and engineering. 
Accuracy considerations are usually taken into account in the 
data processing to reflect the quality of observation and 
measurement procedures, the numerical techniques and related 
operations. The derived quantities are then categorized in terms 
of accuracy and reliability. 
Information usually refers to patterns, features and the like that 
are normally extracted from observation and measurement data 
through processing. Explicit interpretations are not normally 
included in such patterns and tendencies as these can easily be 
context or application dependent. For instance, the estimation of 
a linear or quadratic trend between two data sequences does not 
necessarily include any interpretation of the inferences for the 
variables in question. Hence mathematical information can be 
considered as abstractions or derivations from data without 
including any semantics. Information processing in the 
mathematical sense is logical pattern and similar processing that 
would take any uncertainty or incompleteness of the information 
into consideration. 
Knowledge would then refer to context dependent information 
or interpretations that are common in reasoning like processing. 
In other words, spatial information patterns and trends generally 
have different interpretations and implications for different 
classes of users of the information. For example, a linear or 
quadratic trend between two variables can be the object of 
numerous interpretations. Knowledge processing involves 
facts, rules and procedures that often come from learning 
experience in some specific context or environment. Knowledge 
acquisition, representation and processing are among the 
outstanding research topics in knowledge based system design 
and implementation. 
In terms of abstraction and complexity, information processing 
problems range from data processing algorithmic problems to 
knowledge related questions. The sequel will only consider the 
selection and analysis of appropriate analytical tools and