基于二分类Logistic回归模型的太行山丘陵区县域耕地资源潜力估算

发布时间：2018-03-14 21:12

本文选题：太行山丘陵区　切入点：二分类Logistic回归模型　出处：《中国生态农业学报》2017年04期 　论文类型：期刊论文

【摘要】：耕地红线划定与人-地资源矛盾日益突出背景下,耕地资源潜力的研究与开发日显重要。我国耕地面积近2/3分布在山区,因此山区耕地资源的合理开发利用及其资源潜力的研究尤为重要。本文以华北地区的太行山为研究区域,选择耕地占比和资源潜力最大的丘陵区典型县——河北省井陉县为研究案例,选取13个影响耕地资源潜力的基本生态要素,包括5个地形要素和8个直接气象要素或由气象要素计算得到的间接气象要素,引入二分类Logistic回归分析方法,运用偏最大似然估计向前引入法的拟合方法,筛选提取影响耕地资源潜力的关键生态要素;由模型参数Waldc2统计量分析影响耕地资源潜力的关键生态要素的贡献率排序;由模型参数回归系数β分析耕地资源潜力与生态要素的相关关系;由模型参数发生比率OR分析量化关键生态要素对耕地资源潜力的影响,最终建立Logistic回归模型。基于此模型,在GIS软件中得到井陉县耕地资源潜力分布图,进而估算出县域耕地资源潜力。研究结果表明:13个影响井陉县耕地资源潜力的基本生态要素中8个为关键生态要素;关键生态要素中地形要素配置比气象要素配置更为重要;年平均气温和寒冷指数与耕地资源潜力呈负相关关系,其余生态要素则呈正相关关系;由回归模型估算出井陉县具备垦殖为耕地资源的土地面积为60 400 hm~2,而根据遥感影像解译结果得出的现有耕地资源为45 600 hm~2,由此井陉县尚具有14 800 hm~2的后备耕地资源,相当于现有耕地面积的32.5%,这说明在不考虑垦殖所带来的可能负效应的前提下,井陉县具有较大的后备耕地资源开发潜力,该结论为井陉县后备耕地资源的开发与可持续利用提供了理论依据。
[Abstract]:Under the background of the increasingly prominent contradiction between the delineation of the red line of cultivated land and the human-land resources, the research and development of the potential of cultivated land resources are becoming more and more important. Nearly 2/3 cultivated land areas are distributed in mountainous areas in China. Therefore, it is particularly important to study the rational exploitation and utilization of cultivated land resources and their potential in mountainous areas. This paper takes the Taihang Mountains in North China as the research area. A typical county in hilly region, Jingxing County, Hebei Province, is selected as a case study. Thirteen basic ecological factors affecting the potential of cultivated land resources are selected. Including five topographic elements and eight direct meteorological elements or indirect meteorological elements calculated from meteorological elements, the two-classification Logistic regression analysis method is introduced, and the fitting method of partial maximum likelihood estimation forward introduction method is used. The key ecological factors affecting the potential of cultivated land resources were selected and extracted, and the contribution rates of the key ecological factors affecting the potential of cultivated land resources were analyzed by Waldc2 statistics of model parameters. The correlation between cultivated land resource potential and ecological factors is analyzed by regression coefficient 尾 of model parameters, and the influence of key ecological factors on cultivated land resource potential is quantified by OR of model parameter occurrence ratio. Finally, the Logistic regression model is established based on this model. The potential distribution map of cultivated land resources in Jingxing County was obtained by GIS software, and the potential of cultivated land resources in the county was estimated. The results showed that 8 of the 13 basic ecological factors affecting the potential of cultivated land resources in Jingxing County were key ecological elements. The collocation of topographic elements is more important than that of meteorological elements among the key ecological elements, the annual mean temperature and cold index are negatively correlated with the potential of cultivated land resources, while the other ecological elements are positively correlated. According to the regression model, the area of cultivated land with cultivated land in Jingxing County was estimated to be 60 400 hmm2, while the present cultivated land resource was 45 600 hmm-2 based on the interpretation of remote sensing images. Therefore, Jingxing County still had a reserve cultivated land resource of 14 800 hm~2. It is equivalent to 32.5% of the existing cultivated land area, which shows that Jingxing County has a great potential for the development of reserve cultivated land resources without considering the possible negative effects brought by reclamation. The conclusion provides a theoretical basis for the development and sustainable utilization of the reserve cultivated land resources in Jingxing County.
【作者单位】：中国科学院农业水资源重点实验室/中国科学院遗传与发育生物学研究所农业资源研究中心;中国科学院大学;北京沃尔德防灾绿化技术有限公司;
【基金】：国家重点基础研究发展计划(973计划)项目(2015CB452705)资助~~
【分类号】：F323.211

【相似文献】