Study On Land Use Decision Based On Data Mining

With the speeding up of the industrialization and urbanization in China, the contradiction between people and land is growing due to the population growth and over-exploitation and utilization of resources. Land is the most important resource for a country, and meanwhile, it is the basis for human survival and development. Land use is the core of land problem. In this paper, the land use decision-making is to seek and choose the best land utilization type for achieving the optimal allocation and sustainable use of land resources, improving land use and management level, and ensuring the coordinated development with economic benefits, social benefits and ecological benefits.In this paper, the demand for land use decision-makers was firstly considered. And then a space-time dynamic analysis model of land use decision was building by using multiple spatial data mining method based on the demand of decision-makers. The model was studied at three scales, which include the regional scale, the local scale, and the parcel scale. In this study, the regional scale refers to the entire study area. The local scale is to study the area’s administrative divisions as research subjects. The parcel area refers to the area’s land use status parcels. In this model, land utilization types are measured with land use intensity and land use efficiency. The spatial data mining methods of BP neural network, RBF neural network and wavelet neural network (WNN) were employed. In this paper, land use temporal data mining and decision analysis were applied in Yuzhong Basin. Finally, the choice of land use decision scheme was made to land utilization types, which is divided into 5 types (ecotype, intensive type, transitional type, extensive type, and consumption type), based on game theory. The paper makes a study on the following aspects:(1) This paper compared and analyzed BP neural network, RBF neural network and wavelet neural network (WNN) from theory to application. The models were compared in generalization capability, convergence rate, and error accuracy by using land use data in Yuzhong basin from 2000 to 2008. The results show that WNN model has better fitting effect, stronger generalization, faster convergence, and higher accuracy. (2) At the regional scale, the total dynamic trend of land use in Yuzhong basin from 2000 to 2008 was upward by calculating the index of land use. This shows that from the region as a whole, the index of land use intensity and land use efficiency were growing in Yuzhong basin.(3) At the local scale, land use intensity and land use efficiency were anayzed by using GIS and the three neural network models. The conclusions are as follows:①in agricultural production needs, Jinya is in a semi-intensive used state, Chengguan, Xiaguanying and Xiaogangying are in a semi-extensive used state, and Heping is in a extensive used state;②land utilization type for the construction and development needs in Yuzhong basin is in a extensive transition used state. Jinya is in a semi-intensive used state, Heping is in a semi-extensive used state, and Chengguan, Xiaguanying is in a extensive used state;③in ecological protection needs, Heping and Jinya are in a a semi-extensive used state, meanwhile, Chengguan and Xiaguanying are in a extensive transition used state.(4) At the parcel scale, land use polygons of Yuzhong basin in 2008 as the unit, the land-use spatial decision analysis model was building by using WNN model. It is concluded that①in agricultural production needs, land utilization type in Yuzhong basin is extensive transition type. Cultivated land, garden and facility agriculture land are in a extensive transition use state;②in construction and development needs, land utilization type in Yuzhong basin is also extensive transition type. Urban village and industrial land and transportation land are in a extensive transition use state. And water conservancy facility land is in a ecotype used state;③in ecological protection needs, land utilization type in Yuzhong basin is in a extensive transition used state. Grassland and woodland are also in an extensive transition used state, but waters is in a intensive transition used state.(5) From the perspective of decision makers, the game relationship was analyzed between the decision-making body and decision-making participants in five kinds of land utilization types. The conclusion drawn from the research is that the optimal combination can get the maximum total benefit in three kinds of game relationships, which include government and farmers, government and enterprise, enterprise and farmers. The optimal combination could not only improve land use efficiency, but also have no obstacle driving from farmers and enterprise. Moreover, although the game cost has been increased to both sides, the fifth strategy combination could also achieve a balanced game relationship.