高会, 谭莉梅, 刘鹏, 刘金铜, 李晓荣. 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[J]. 中国生态农业学报(中英文), 2017, 25(4): 490-497. DOI: 10.13930/j.cnki.cjea.160778
引用本文: 高会, 谭莉梅, 刘鹏, 刘金铜, 李晓荣. 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[J]. 中国生态农业学报(中英文), 2017, 25(4): 490-497. DOI: 10.13930/j.cnki.cjea.160778
GAO Hui, TAN Limei, LIU Peng, LIU Jintong, LI Xiaorong. Estimation of arable land resources potential in hilly area of Taihang Mountain based on binary Logistic regression model[J]. Chinese Journal of Eco-Agriculture, 2017, 25(4): 490-497. DOI: 10.13930/j.cnki.cjea.160778
Citation: GAO Hui, TAN Limei, LIU Peng, LIU Jintong, LI Xiaorong. Estimation of arable land resources potential in hilly area of Taihang Mountain based on binary Logistic regression model[J]. Chinese Journal of Eco-Agriculture, 2017, 25(4): 490-497. DOI: 10.13930/j.cnki.cjea.160778

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

Estimation of arable land resources potential in hilly area of Taihang Mountain based on binary Logistic regression model

  • 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Waldχ2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 hm2,而根据遥感影像解译结果得出的现有耕地资源为45 600 hm2,由此井陉县尚具有14 800 hm2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: Research on the potential of cultivated land resources in mountainous area is particularly important for the sustainable development and utilization of arable lands. Using Jingxing County of Hebei Province, a typical hilly county in Taihang Mountain, as a case study, the potential for reserved cultivated land in mountain regions was estimated. A total of 13 ecological factors influencing the potential of cultivated land resources in hilly areas were used in the analysis-5 terrain factors and 8 climate or climate-related factors. A binary Logistic regression model was built on key ecological factors to analyze the potential of arable land resources in mountain regions in the case-study area. The distribution map of potential arable land resources in Jingxing County was draw in GIS environment based on the result binary Logistic regress and the potential arable land resources in the region analyzed from the map. The key ecological factors among 13 factors that influenced arable land resources were extracted using the same binary Logistic regression model. The order of the contribution of key ecological factors was determined using the model parameter (Waldχ2). The regression coefficient (β) was used to analyze the correlation between the key ecological factors and potential arable land resources in the region. Odds ratio (OR) was use to show correlation between the changes in configuration of key ecological factors and the changes in potential arable land resources. The results showed that there were 8 key ecological factors influencing cultivated land resources in the case-study area, with the importance order of relief > elevation > slope position index > trasp > coldness index > annual average temperature > drought index > slope. Terrain factors were more important indicators than the climate-related factors in estimation of cultivated land resource potential. Among 8 key ecological factors, annual average temperature and coldness index were both negatively and positively correlated with potential arable land resources, while other 6 factors were all negatively correlated with potential arable land resources. OR calculation showed that except drought index, the one unit change of the key ecological factors caused 1 time change in cultivated land resource potential. About 60 400 hm2 of land resources were available for cultivation in the Jingxing County. The analysis of SPOT5 image showed 45 600 hm2 of existing cultivated land. Therefore, there were 14 800 hm2 of cultivatable land reserves in the county, accounted for 32.5% of the existing cultivatable lands in the region. This suggested that there still were a lot of cultivatable land reserves in Jingxing County. The results of this study provided theory basis for the development and utilization of cultivated land reserves in Jingxing County.

     

/

返回文章
返回