THE VISUALISATION OF GIS GENERATED INFORMATION QUALITY 
Paper #1258, Technical Session 26, Thursday August 6th. 1992 
Corne P.J.M. van Elzakker 
International Institute for Aerospace Survey and Earth Sciences (ITC), The Netherlands 
Bheshem Ramlal 
University of the West Indies, St. Augustine, Trinidad and Tobago 
Jane E. Drummond 
International Institute for Aerospace Survey and Earth Sciences (ITC), The Netherlands 
PURPOSE: 
Many centuries of tradition and research have contributed to the design of paper maps, and some decades 
of effort have gone into the design of the screen displays used in GIS. Recently consideration has been 
given to the error propagated in a GIS as it generates new information but there are no standardised 
designs for displaying the quality of that information. This paper examines some possible means of 
displaying information quality, and describes such means now implemented in ILVIS, a PC based GIS. 
Examples are based on a Land Reallocation project where, in particular, topographic and soils data sets 
were processed to provide both information needed for reallocation and the quality of that reallocation 
information. 
KEY WORDS: GIS, Data Quality, Information Quality, Cartographic Design 
Priciples, Graphic Display, Error Propagation in the GIS environment. 
1. INTRODUCTION 
Although photogrammetrists and other geodesists 
have long been concerned with the quality of the 
spatial information gathered by their systems, 
only recently has data quality within GIS become a 
"hot topic" - as evidenced in the 1989 publication 
of Goodchild and Gopal’s "Accuracy of Spatial 
Databases", NCGIA (the United States National 
Centre for Geographic Information and Analysis) 
support for comprehensive reviews of data quality 
[VEREGIN, 1989] and initial attempts to formalize 
its visualisation [CLAPMAN and BEARD,1991], and 
numerous degree theses. This recent GIS-centred 
activity seems to have been started by Chrisman 
[CHRISMAN, 1982] and Blakemore  [BLAKEMORE, 1984] 
in the early 1980’s, with their popularisation of 
Perkal’s "epsilon band" concept. However concerns 
over the quality of digital land-use data when 
derived from satellite remote sensing sources 
generated approaches in the 1970's (e.g. [HORD and 
BROONER, 1976] and [VAN GENDEREN and LOCK, 1977]), 
which now (in the 1990’s, as raster based display 
media become the ’norm’) are increasingly applied 
(e.g. [BURROUGH and HEUVELINK, 1992]). A result of 
recent NCGIA and US National Committee for Digital 
Cartographic Data Standards activities [NCDCDS, 
1988] is the current acceptance of a five part 
division of the problems of geographic (spatial) 
data and information quality into: spatial 
quality; attribute quality; completeness; logical 
consistency; and lineage. Although other 
subdivisions (e.g. [RADWAN, SUHARTO and SUTRISNO 
YONO, 1991]) may be more manageable, in the 
investigation reported here the well publicised 
fivefold classification will be used. 
One standard 'pre-GIS' technique for the display 
of data quality and popular with land-surveyors 
has been to use ‘error ellipses’ (e.g. [RICHARDUS, 
1974]) whereas amongst photogrammetrists 
discrepancy vectors have been used (e.g. [ASPRS, 
1980]), but, despite the popularisation of GIS and 
the recent interest in GIS data and information 
quality, no other standard techniques for 
displaying the quality of that data and 
information have been widely applied. Error 
ellipses and discrepancy vectors are probably only 
608 
suitable for displaying the quality of the spatial 
information of a very few points at a time, and 
for some aspects of that information only. Based 
on established cartographic theory, this paper 
examines some possible means of displaying GIS 
generated information quality at different 
measurement levels, and describes such means now 
being implemented in ILVIS, a PC based GIS. 
Examples are based on a Dutch Land Reallocation 
project where, in particular, topographic and 
soils data sets were processed to provide both 
information needed for reallocation and the 
quality of that reallocation information. 
The investigation reported on here will contribute 
to the eventual completion of a GIS tool handling 
information quality which will be termed the 
"Uncertainty Subsystem" of ILWIS. It is hoped it 
will process quality information in parallel with 
the information generated for the users’ 
applications, and provide the resulting quality 
information at the user’s request. 
This paper will first examine present cartographic 
applications of graphic semiology, which should 
identify the most appropriate methods for 
displaying quality information. Then, after a 
brief description of ILWIS, an investigation of 
mainly the cartographic aspects of the (still) 
prototype "Uncertainty Subsystem" within ILWIS 
using a land reallocation case study will be 
described. (Note: further details of the 
theoretical basis of the "Uncertainty Subsystem" 
are provided in an ISPRS Congress XXVII Commission 
III presentation "A GIS Uncertainty Subsystem" 
[RAMLAL & DRUMMOND, 1992] also published in these 
Archives.) 
2. THE CARTOGRAPHIC REPRESENTATION OF 
QUALITY INFORMATION 
We accept that hardcopy or softcopy maps are the 
most efficient means for communicating geographic 
(or spatial) information, thus whenever quality 
information is also of a geographic nature, it 
ede F de D) (C) de (C) «7 (CT 0C) 8) MO MH ID 
NO 
ct OC rt à — tmo HB "tj nn X3 ct.O (D oo UD o£ CY be (OL KO) 0O O xr 
ct t O ct 0 0 Ht 
Hd OH OH DAN