近20年中国耕地数量变化趋势及其驱动因子分析

RESEARCH ON THE CHANGE TREND OF FARMLAND QUANTITY IN CHINA FOR RECENT 20 YEARS AND ITS DRIVING FACTORS

关键词:  数据挖掘ARIMA模型耕地数量变化趋势驱动因子

Key words:  data mining  ARIMA model  farmland quantity  change trend  driving factors

摘要:

［目的］研究“二调”前后近20年我国耕地面积的变化趋势及其影响因子，为促进耕地保护和土地资源的高效集约节约利用提供决策支撑。［方法］基于1996—2016年耕地数量的原始统计数据，利用数据挖掘技术选用ARIMA模型，基于两次全国土地调查时间段耕地数量数据进行相互预测和反推，即利用1996—2008年耕地数量数据预测2009年耕地数量，利用2009—2016年数据反推2008年耕地数量，从而实现数据的校正和比值归一化的目的，进一步研究近20年内我国耕地数量的变化趋势，并通过相关性分析研究遴选出其驱动因子。［结果］趋势研究结果表明，近20年内我国耕地数量呈现逐年递减的趋势，其中2004年之前递减速度较快，之后递减速度趋缓。相关性分析结果显示，第一产值增加值占比、年末总人口数，以及城镇人口数是影响耕地数量变化的重要驱动因子，其相关系数分别为0959，-0918，-0896。［结论］从研究结果可以推论，我国在社会经济发展中的产业结构布局，以及城镇化建设进程对耕地数量变化产生了重要的影响。

Abstract:

The research aims to research the changing trend of farmland quantity for recent 20 years before and after the second national land use survey. It could provide decision support for farmland protection and efficient use of land resources. The original data of farmland area from 1996 to 2016 were collected, data mining technology was introduced, and ARIMA model was chosen to predict the farmland quantity each other based on the farmland area of the twice land use survey, which meant to predict the farmland quantity of 2009 with the farmland quantity of the years from 1996 to 2008, and to predict the farmland quantity of 2008 inversely with the farmland quantity of the years from 2009 to 2016, so the original farmland area data could be corrected and normalized. Then the changing trend of farmland quantity for recent 20 years was studied furtherly and correlation analysis was used to study the driving factors. The trend research result showed, the farmland area decreased year by year for recent 20 years, the decreasing rate was faster before 2004, and then it decreased slowly. The correlation analysis result showed, the added value proportion of primary industry, the total population, and the urban population were the most important driving factors which influenced the farmland quantity change, their coefficients were 0.959, -0.918 and -0.896 respectively. It could be concluded by the result that the industry configuration and the urbanization could affect the farmland quantity with the development of social economy.