Study On Cultivated Land Area Change And Its Driving Forces In Hunan Province

Arable land resources is the production of human survival, is the most important guarantee of the development of food production. Analysis of arable land change driving factor, the trend of forecast arable land, is conducive to the rational allocation of arable land resources, to protect the safety of food production, to promote the harmonious development of society.Hunan Province is the most important grain production base in central China. As a major agricultural province, and the change of its arable land is closely related with the Hunan Province and even the country’s food security problem. Since the reform and opening up, especially since1990, the country has developed after a series of policies and measures to accelerate the pace of development in the central and western regions, the rapid economic growth in Hunan Province, cities and towns are rapidly spreading to rural areas, building land occupation of cultivated land resources is increasing, fewer people and more contradictory deteriorating. Analysis of Cultivated Land area change impact factor to accurately predict the change in trend of the cultivated area in Hunan Province, and is of positive significance to the development of a sustainable development strategy in line with the actual situation of Cultivated Land.The impact of arable land change driving factor complex and changing, it is difficult to determine. At present, the principal component analysis, correlation analysis, gray correlation degree method to analyze the driving factors of the cultivated area. However, these methods are a lot of deficiencies, still there is a lot of information overlap between selected factors, it is difficult to guarantee the accuracy of prediction of the arable land. Arable land change data is a complex nonlinear time-series data, BP neural network prediction method (back-propagation neural network, BPNN) model, support vector regression (support vector machine regression, SVR) model and least squares support vector regression (least squares support vector machine, LSSVM) model. These predictive models are nonlinear system model, forecast of arable land has reached a certain degree of accuracy. Arable land also has significant characteristics of time series, simply use the traditional model its predictions fall far short of the actual production needs, they must bring the relevant time series analysis techniques in order to further improve the prediction accuracy of the arable land. GS-SVR model is a combination of geostatistics (geo-statistics, GS) model order principle and a high precision of the SVR model time series forecasting methods.In this paper, the argument based on the model of the GS-SVR full combination forecasting mean square error (Mean Squared Error, MSE) minimum principle to determine the optimal combination of driving factors of arable land changes. In Hunan1978-1999arable land change data, for example, first seven major driving factors extracted by principal component analysis and correlation analysis, and then through the GS-SVR argument combination forecasting MSE minimum principle to determine the optimal combination of impact factor level of urbanization and the real estate industry output value index, and further by the Hunan Province2000-2008arable land change data, for example, to verify the effectiveness of the selected drive factor combinations. The article also commonly used by the SVR-CAR, LSSVM, BPNN, ARIMA and MLRR time series prediction method to verify the effectiveness of the selected driving factor combinations. The results showed that GS-SVR combination selected combination of driving factors greatly improve a variety of commonly used time series methods to predict the accuracy of the cultivated area, the level of urbanization and the real estate industry output value of the index is the main driving factor affecting the change of cultivated area in Hunan Province. In addition, the highest prediction accuracy of the model GS-SVR method, GS-SVR is an effective way to predict the change of arable land, can provide a useful reference for the arable land area researchers.