基于可耕性评价与多目标权衡的耕地布局优化  

Optimizing cultivated land layout using cultivability evaluation and multi-objective trade-offs

作　　者：黄炳元 黄秋昊[1,3,4,5] 阳艳 郑锦浩 陈逸航 HUANG Bingyuan;HUANG Qiuhao;YANG Yan;ZHENG Jinhao;CHEN Yihang(School of Geography and Ocean Science,Nanjing University,Nanjing 210023,China;Key Laboratory of the Coastal Zone Exploitation and Protection,Ministry of Natural Resources,Nanjing 210023,China;Key Laboratory for Land Satellite Remote Sensing Applications,Ministry of Natural Resources,Nanjing 210023,China;Jiangsu Provincial Key Laboratory of Geographic Information Science and Technology,Nanjing 210023,China;Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application,Nanjing 210023,China)

机构地区：[1]南京大学地理与海洋科学学院,南京210023 [2]自然资源部海岸带开发与保护重点实验室,南京210023 [3]自然资源部国土卫星遥感应用重点实验室,南京210023 [4]江苏省地理信息技术重点实验室,南京210023 [5]江苏省地理信息资源开发与协同创新中心,南京210023

出　　处：《农业工程学报》2024年第19期240-249,共10页Transactions of the Chinese Society of Agricultural Engineering

基　　金：国家自然科学基金面上项目(41571082)。

摘　　要：为探索各种空间因素作用下耕地未来空间布局的发展方向及其管控策略,该研究以位于长三角地区的常州市为例,采用机器学习算法构建了土地资源可耕性评价模型,并分别用PLUS模型和InVEST模型计算城市发展潜力和生境保护潜力,最后使用蚁群优化算法实现权衡多目标的耕地空间配置模拟并划定优化分区。结果表明:1)通过土地资源可耕性评价发现,常州市不可耕作区中有168.25 km^(2)现状为耕地,高度可耕作区中有254.11 km^(2)为非耕地,耕地资源分布存在明显的不均衡性。2)对比不同情景下的耕地空间配置结果,确定城市发展-生境保护-集中耕作情景取得了最理想的效果,平均可耕作潜力为0.9387,有助于优化耕地资源合理布局。3)统筹耕地空间配置结果与土地利用现状将常州市划分为核心保护区、质量提升区、潜力储备区、建设缓冲区、生态保育区共5区并提出差异化的优化管控策略。该研究的技术路径和结果对于重新认识区域耕地资源本底、有效调节土地资源错配具有参考意义。Here future spatial layout of cultivated land was investigated under the influence of various spatial factors in the city of Changzhou in the Yangtze River Delta.The cultivability of land resource was evaluated to integrate the urban development potential,habitat conservation potential,and the degree of agglomeration.A machine learning-based model was constructed to assess the land cultivability using multiple algorithms.CatBoost also demonstrated the highest accuracy of classification.The PLUS and InVEST model were then used to calculate the urban development and habitat conservation potential,respectively.Finally,the resulting sub-objectives were input into an Ant Colony Optimization(ACO).As such,the multi-objective spatial allocation of cultivated land was simulated to delineate the optimization zones.The results show that:1)Training labels were set using NPP(Net Primary Productivity),elevation,and slope in the land cultivability assessment,considering the influencing factors,such as topography,soil,climate,location,and irrigation.CatBoost was outperformed the rest in the accuracy of classification.The evaluation of cultivability revealed that Changzhou shared a potential cultivated land area exceeding the current cultivated land by 543.61 km^(2),indicating the potential for further cultivation.However,168.25 km²of the current cultivated land was located in uncultivatable zones,while 254.11 km^(2) of highly cultivatable land remained unused,indicating the significantly spatial imbalances.2)Changzhou's urban development and habitat conservation potential were calculated using the PLUS and InVEST models,respectively,in order to manage land use conflicts during the optimization of cultivated land layout.The urban development potential exhibited both central outward expansion and road-oriented diffusion.The habitat conservation potential showed the high potential in water bodies and forests,while the low potential in built-up areas.Different weight values were set for the sub-objectives in the utility function.Th

关 键 词：耕地布局优化 多目标权衡 机器学习 蚁群优化算法 常州市 

分 类 号：F301.21[经济管理—产业经济]