| 摘要: |
| ［目的］分析2000—2018年中国耕地利用变化的文献，对学者在研究耕地利用变化驱动和模拟方面所使用的方法进行对比分析，系统梳理各种方法在应用上存在的优势与不足，为未来学者开展相关研究提供方法参考。［方法］文献分析法、分类统计法等。［结果］对于耕地利用变化驱动研究而言，与主成分分析法组合的方法使用频率较高，基本适用于所有尺度，该方法可使复杂系统简单化、提高精度、有利于确定主要驱动因子，但一些重要的驱动因子难以经过统计学检验进入回归模型。对于耕地利用变化模拟研究而言，BP神经网络模型适合于大尺度的模拟研究，该方法预测能力强但精度一般； 灰色动态模型GM（1， 1）进行模拟所需的数据量较少、精度高，但受随机因素影响较大。CLUE-S模型已经比较成熟，在中小尺度上得到了很好的应用。［结论］耕地利用变化驱动与模拟的研究方法众多，但使用时相对单一且方法适用尺度的局限性较大。耕地利用变化驱动研究中学者们更倾向于使用定性与定量相结合的研究方法，未来耕地利用变化驱动的研究方法将逐渐朝着全定量的方向发展。耕地利用变化模拟研究中学者们更倾向于单一模型的使用，未来综合模型跨尺度融合将成为耕地利用变化模拟研究方法的发展方向。总体来看，未来耕地利用变化驱动与模拟研究方法将朝着双向、动态的复合型研究方法方向发展。 |
| 关键词: 土地资源方法研究文献综述耕地土地利用驱动力模拟 |
| DOI： |
| 分类号:F301 |
| 基金项目:教育部人文社科基金项目“三峡库区快速城镇化地区耕地利用管理对策研究”（15YJC630005）； 北京市社会科学基金“京津冀潮白河流域耕地变化驱动力与协同管理对策”（17GLC063） |
|
| METHODS ANALYSIS ON CULTIVATED LAND USE CHANGES IN CHINA*——BASED ON DRIVING AND SIMULATION |
|
Su Ruiqing1, Cao Yingui1,2※
|
|
1.School of Land Science and Technology, China University of Geosciences, Beijing 100083, China;2. Key Lab of Land Consolidation, Ministry of Nature Resources of the PRC, Beijing 100035, China
|
| Abstract: |
| By analyzing the literature of cultivated land use change in China from 2000 to 2018, this research made a comparative analysis on the methods used by scholars, and systematically sorted out the advantages and disadvantages of various methods in the application of cultivated land use change, which was beneficial for scholars to grasp the advantages and disadvantages of different methods, and so as to provide reference for the future researches on the changes of cultivated land use. The literature analysis method and the classification statistical method were adopted in the process of research. The results were showed as follows. For the driving of cultivated land use change, the method combined with principal component analysis, which was basically applicable to all scales and widely used in the study of the driving forces of cultivated land use change. The principal component analysis could simplify the complex systems, improve the precision and help to determine the main driving factors, but some important driving factors were difficult to enter the regression model through statistical tests. For the simulation study of cultivated land use change, many methods had a good application. The Back Propagation neural network model was suitable for large scale simulation research, and it had strong prediction ability but general accuracy. The grey dynamic model GM (1, 1) required less data and had high precision for simulation, but it was greatly influenced by random factors. The CLUE S model had already been mature and well applied in small and medium scales. We found that there were many driving forces and simulation methods to study the change of cultivated land use, and these methods had their own advantages and limitations. Many methods were used separately and limited in scale. In the study of driving forces of cultivated land use change, scholars preferred to use qualitative and quantitative research methods. In the future, the research methods of the driving of cultivated land use change will gradually develop towards the direction of full quantification. In the study of simulation of cultivated land use change, scholars preferred to use a single model. In the future, development direction of research method on simulation of cultivated land use change will be cross scale fusion of comprehensive models. In general, the research methods of the driving and simulation methods of cultivated land use change will develop toward bidirectional and dynamic compound research methods in the future. |
| Key words: land resource methods research literature analysis cultivated land land use driving forces simulation |
