Uncertainty of spatial data in GIS can be in the aspect of position

attribute

temporary

logical relationship and completeness. Among them

attribute uncertainty can directly affect the quality of GIS-based decision making. In this paper

the concept and propagation of the attribute uncertainty are firstly described. Then the relationship between the attribute and position uncertainties is analyzed. Moreover

the theories and methods to study and visualize the attribute uncertainty are discussed. Attribute uncertainty of spatial describes the difference between the observed values of an entity attribute with their true values in spatial and temporal space. There are a number of theories and developed methods to deal with attribute uncertainties. These may include

for example

(1) object-and filed-based model; (2) probability theory

evidence theory

spatial statistics

"S" band model and probability vector which deal with random attribute uncertainties

while fuzzy set

rough set

genetic algorithms which deal with imprecise attribute uncertainties. Cloud theory can be used to deal with both randomness and fuzziness of attribute uncertainty. The relationship betweenxandμ(x)

fuzzy set is one to one

rough set is one to a region

and cloud theory is on one to a piece of cloud; (4) Visualization of attribute uncertainty which is based on Bertin geographical parameters and their extensions. Three-dimensional visualization and virtual reality are also helpful. Finally

the future research issues on attribute uncertainty are given.