ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001 
402 
which accounts for 76% of accuracy. Conventional soil maps 
typically achieved 60-70% of accuracy. The 50 sites can be 
divided into two groups: one on the drumlin areas and the 
other on the inter-drumlin areas. The accuracy for the sites in 
the drumlin areas reaches 90% while the accuracy for the 
inter-drumlin areas is about 63%. The difference in accuracies 
between the two areas is due to the difference in 
environmental gradients and the fact that SoLIM is sensitive to 
the ability of characterizing soil formative environmental 
conditions. The relief (gradient) in the drumlin areas is much 
stronger than that over the inter-drumlin areas. For the drumlin 
areas, the soil formative environment conditions for different 
soils can be easily distinguished (characterized) using GIS due 
to the stronger relief. As a result, the soil-landscape model is 
more ‘faithful” applied through SoLIM. However, the soil 
formative environment conditions for different soils in the inter- 
drumlin areas is very difficult to distinguish due to low relief 
and it is difficult to distinguish soils over these areas using the 
environment conditions. Overall, we believe that the soil- 
landscape model developed through the use of FCM 
methodology is of good quality since the accuracy of the soil 
map exceeds that of common soil maps produced through 
extensive soil surveys. 
Legend 
Kidder 
McHenry 
St.Charles 
Mayville 
Virgil 
Sable 
Figure 2: Soil map produced from SoLIM using the soil-landscape model constructed using the FCM-based method. 
4. Summary 
This paper presents a methodology to assist the development 
of knowledge of relationships between a given geographic 
phenomenon and its environmental conditions. The method 
employed a fuzzy c-means classification to identify unique 
combinations of environmental conditions and to discern 
locations of these unique combinations. The results (the 
unique combinations and the spatial locations of these unique 
combinations) were then used to direct field investigation 
efforts and to improve the efficiency of acquisition of 
knowledge on the relationships in the field. 
Through a soil mapping case study it was found that the FCM 
assisted field investigation was effective in developing the 
knowledge of the soil-environmental relationships. First, the 
amount of field observations was reduced. Second, the 
acquired knowledge of the relationships was of high quality. 
Acknowledgements 
The research reported here was supported by a research grant 
from the Graduate School, University of Wisconsin-Madison, 
by the funding from Natural Resource Conservation Service, 
United States Department of Agriculture under Agreement No. 
69-5F48-9-00186. The State Key Lab of Resources and 
Environmental Information System (LREIS), Institute of 
Geography, Chinese Academy of Sciences kindly provided 
travel support to the senior author to attend this meeting. 
References 
Ahn, C. W., M. F. Baumgardner, and L. L. Biehl. 1999. 
Delineation of soil variability using geostatistics and fuzzy 
clustering analyses of hyperspectral data. Soil Science 
Society of America Journal 63, 142 - 150. 
Bezdek, J.C. 1981. Pattern Recognition with Fuzzy Objective 
Function Algorithms. New York: Plenum Press. 
Bezdek, J. C., R. Ehrlich, W. Full. 1984. FCM: The fuzzy c- 
means clustering algorithm. Computers and Geosciences 
10 (2-3), 191 -203. 
McSweeney, K., B.K. Slater, R.D. Hammer, J.C. Bell, P.E. 
Gessler, and G.W. Petersen. Towards a new framework