Research On Interval Uncertainty In Spatial Position Data Of GIS

The quality and uncertainty of spatial databases is one of the most major problems.Using probability theory and mathematical statistics which is the main study method, theresearch has achieved great developments. However, much of researches ignore the intervaluncertainty. While the objective world is complex with uncertainty, human cognition is alsolimited with fuzziness. It is hard to obtain exact information knowledge about observedobjects, but can get their interval ranges. These interval uncertainties would be propagatedand accumulated along with the GIS operations, finally has an important effect on the resultsof decision analysis. It is necessary to further analysis and discuss.In order to deal with interval uncertainty, this paper starts by introducing the intervalanalysis ways, studies uncertainty in the fundamental functions of GIS, and focus on themethod of interval uncertainty propagation which is based on Taylor-series expansion. Thennew distance measures between two interval data are proposed and applied to fuzzy c-meansclustering algorithm.The main works of this thesis are as follows:(1) We introduce three ways to manage interval uncertainty. Using them to calculate thelength of line, the interaction of line and the area of polygon, it shows that Taylor-seriesexpansion method and interval slope extension method are feasible.(2) Taylor series expansion offers theoretic support to study uncertainty propagation ofspatial interval data. It provides propagation function and is given correspondingly andvalidated by simulation experiments.(3) After listing four existent distance definitions, this paper presents two new methods.A comparative study is implemented on different fuzzy c-means clustering algorithms to varydistance measures. It turns out that new algorithmsachieve better results and attain a sameeffect in the whole.