GRUIJTER, 1986] it was found that the standard 
deviation of Moisture Supply determinations is 
17mm. With this information, and using estimation 
by confidence intervals the pro- bability (e.g.) 
of a landparcel having Moisture Supply Capacity 
Class 2, when its Moisture Supply Capacity has 
been measured to be 166mm is 81%. 
4.5 Quality of the Grazing Suitability 
Classification 
Taking into account the quality of the model (see 
section 4.1), the quality of the Soil Drainage 
Status Level (section 4.2), the Soil Bearing 
Capacity Class (section 4.3), the Moisture Supply 
Capacity Class (section 4.4), and using Crisp Set 
Theory it is possible to estimate the probability 
of the given landparcel (referred to in sections 
4.2, 4.3, 4.4) having the predicted Grazing 
Suitability to be: 
P = 0.98(0.85 *x 0.82 * 0.81) = .55,= 55% 
Applying Fuzzy Sub-Set Theory [KAUFMANN, 1975] and 
using these probabilities as Certainty Factors, 
the overall Certainty Factor associated with the 
predicted Grazing Suitability would be 0.81. 
It is such probabilities or certainty factors 
which may be displayed, along with grazing 
suitability either by cartographic or other means, 
to provide the GIS user with information on the 
quality of the generated information. 
4.6 Results of the exploration of the Land 
Reallocation Model 
In this study a database was built in ILWIS which 
held the land parcel boundaries supplied by the 
Dutch Topographic Service, Soil Polygons supplied 
by the Dutch Soil Research Institute, and database 
tables holding the soil characteristics and the 
relevant soil characteristics quality parameters 
of the those soil polygons. 
First using the available ILVIS facilities and 
selecting a low-cost ink-jet plotter as output 
device a map showing just the quality of the soils 
data was produced, in 4 classes represented by 
means of the visual variable value (Figure 6 ). 
Then using the same ILWIS facilities the Grazing 
Suitability Model was inserted and a multicoloured 
5-class grazing suitability map produced. 
Thereafter using the procedures outlined in 
Sections 4.1 to 4.4 and implemented in ILWIS the 
quality parameters were processed to give i) a 
2-class probability map (<50% probability, >50% 
probability); ii) a 3-class probability map (low, 
average, and good probability); and iii) a 5-class 
probability map (<10%, 10-30%, 30-40%, 40-50%, and 
50-60%). The 3-class map is shown in Figure 9 . 
The probability information represented in this 
FIGURE 11 was then combined with the grazing 
suitability information as shown in the 
multicoloured suitability map referred to above. 
As the visual variable value had to be reserved 
for the representation of the (ordered) 
suitability information already, data quality 
could not be shown by varying the relative 
lightness or darkness of the colours of the 
suitability classes. The solution selected was a 
coarse grey stipple overlay of three desity 
classes corresponding to the probability classes. 
614 
5. CONCLUSIONS 
A team at ITC is continuing to work on developing 
this "Uncertainty Subsystem". This includes 
Cartography staffmembers with an interest in 
graphic semiology and the optimization of 
soft-copy display in a GIS environment, as well as 
students who are now concentrating on other 
aspects of the subsystem - including error 
propogation in dynamic diffusion models relating 
to industrial hazards, and developing a 
user-friendly interface for variance propagation 
in any mathematical processing models. We aim to 
have the ILVIS "Uncertainty Subsystem" completed 
by the end of 1993. 
6. ACKNOWLEDGEMENTS 
We are grateful to the financial support provided 
by the Directorate General of International 
Cooperation of the Ministry of Foreign Affairs 
(Government of the Netherlands); the scientific 
support provided the Winand Staring Centrum (Dutch 
Soil Research Institute), Landinrichtingsdienst 
(Dutch Land Reallocation Service of the Ministry 
of Agriculture, Land Management, and Fishery), 
Topografische Dienst (Dutch Topographic Service); 
and support provided by Mr. Wim Feringa and Mr. 
Henk A.W. Scholten of ITC. 
7. REFERENCES 
ASPRS, 1980 "Manual of Photogrammetry", American 
Society of Photogrammetry and Remote Sensing, 
Falls; Church, Virginia, .USA, ed. Slama, C.C., 
p338. 
Bertin, J. (1981), Graphics and graphic 
information processing. Berlin and New York: 
Walter de Gruyter. 
Bertin, J. (1983), Semiology of graphics. Madison: 
University of Wisconsin Press. Translated by W.J. 
Berg. 
Blakemore, M., 1984 "Generalization and Error in 
Spatial Databases" Cartografica, Vol. 21. 
Bos, E.S. (1984), Systematic symbol design in 
cartographic education. ITC Journal 1984-1, pp. 
20-28. 
Burrough,P.A., and Heuvelink, G.B.M., 1992 "The 
Sensitivity of Boolean and Continuous (Fuzzy) 
Logical Modelling to Uncertain Data", Proceeedings 
EGIS ’92 Third European Conference and Exhibition 
on Geographical Information Systems", Munich, 
Germany, Vol. 2, pp1032-1041. 
Chrisman, N., 1982 "A Theory of Cartographic Error 
and its Measurement in Digital Databases", 
Proceedings AutoCarto 5 1982. 
Chrisman, N., and McGranaghan, M., 1990 "Accuracy 
of Spatial Databases", Unit 45 in Technical Isuues 
of GIS of the NCGIA Core Curriculum 1990. 
Clapham, S.B., and Beard, EK.» 1991 "The 
Development of an Initial Framework for the 
Visualization of Spatial Data Quality", Technical 
Papers 1991 ACSM-ASPRS Annual Convention 
Baltimore, USA, Vol. 2, pp73-82, (Cartography and 
GIS/LIS). 
PS P^ A za