With the development of "3S" technologies, a large quantity of spatial-temporal data related to land use has been
accessed. Being scattered across different departments and lacking of relevant analysis tools made them utilize
insufficiently. Although some experts have applied data mining to solve this problem, most of them have only provided
one method for single task to build the mining systems. However, it is undesirable to use just one method to mine. In
addition, the single function systems can not be used widely and conveniently. Hence, under full investigation on
operations of land use, a multi-source data mining prototype system for land use is proposed by integrating of
technologies of GIS and spatial data mining. According to the general data mining process, aiming at the multi-demands
of land evaluation and land planning and so on, the system is developed by using ArcEngine 9.0 and VB.net. The system
integrates basic geospatial data, land use/cover data, and thematic data as data sources, excavates different knowledge of
land Quality, land use zoning rules, land use patterns and change rules and so on. Based on the types of knowledge, the
system accordingly provides several different mining methods, including decision tree, support vector machine, artificial
neural network, time series, spatial association rules, etc. Wide adaptability of the system is demonstrated by using some
cases. The results of the system can meet multipurpose needs and be used to support decision-making of the land
management department.
|