Research On The Prediction Model Of Land Area Forecasting And Land Layout Simulation

Land resource is one of the important basis of city. As the city sprawling, the contradictions between the limited land resource and the increasing demand of land and between exploding population and the developing economy are becoming more and more significantly. Therefore, forecasting land use changing efficiently not only plays an important role in the development of national economy and the improvement of people's living quality but also provides a scientific reference for the overall planning of land use.Most researches on land use change can be classified into two methods:one is land area forecasting based on statistical model; the other is the simulation of land layout using GIS technology. More specifically, with regards to land area forecasting, the statistical models can be divided into regression models and multiple regression models according to the number of parameters and the single forecasting models and combined forecasting models according to the number of models, respectively. On the other hand, for the simulation of land layout, the methods of land layout arrangement includes cellular automata model, multi-agent system, a system dynamics model, etc.; according to different spatial algorithms, the models can be divided into intelligent space algorithm and non intelligent space algorithm. This paper examines the performance of various forecast and simulation models working in land use change, and designs a land area hybrid predictive model based on land status and a land layout simulation CA model with Markov-C5.0classification model. Furthermore, a framework which can be modified dynamically based on these two models is developed, of forecast and simulation the land use change dynamically.This paper reviews the last three land use classification standards enacted in2007,2002and1984in China. According to the demand of research to re-class the land into nine types: urban land, town land, village land, mining land, scenic spot land, agricultu re and forestry land, transportation land,water and conservancy facilities land and the other land. Based on classing driving factors into natural, social and economic, policies and the other factors, describe the impacts about land use change of these factors by qualitative analysis and quantitative analysis. Then modeling the nine types of land area forecast hybrid models using correlation analysis and principal component analysis, including population, non-agricultural population, agricultural population, total GDP, the first industry GDP, the second and third industry GDP, the first industry investment, the second and third industry investment, car ownership, per road area, road density, park area, green rate, the rate per capita.With the regards to the land area forecasting, this paper proposes a hybrid model of different land types. This model is based on the hybrid model framework, and uses the best regression and multiple factor regression model of each land type to create the predictive model. The researchers can understand the relationship between different lands and all kinds of driving factors and the inner relationship of each kinds of land by the hybrid model.For the land layout simulation, this paper proposes a new simulation model with Markov-C5.0classification algorithm including three stages. The first stage is to obtain land circulating direction by Markov model, and then to calculate the growth of land by the land area hybrid predictive model. The second stage is to fuse the space factors by principal component analysis. The third stage is to obtain the conversion rules of land by C5.0classification algorithm. In addition, analyzing the dynamic and scheduling in land area forecasting and land layout simulation, and dividing the dynamics into the process of land scale temporal, spatial dynamic factors and control factor dynamic. A model framework is designed to update these dynamics, and make the CA model can be automatic learning and real-time adjustment.Finally, forecast and simulate the land area and layout of Wuhan in2025. The result shows that the model framework designed in this paper is advanced and availability. This model can work out the land structure and distribution efficiently, and the correction model framework can enhance the usability of the ex-prediction results. In addition, using mathematical and statistics model, GIS spatial analysis model and algorithm, ArcGIS development keys, a land area forecast and land layout simulation system is developed, which evaluates the outcomes of this research.