昆明市土地利用变化趋势的多模型对比分析  

Comparative Analysis of Three Land Use Transition Models in Kunming City

作　　者：许安泽 张述清[2] 朱大明[1] 裘木兰 保红[3] Xu Anze;Zhang Shuqing;Zhu Daming;Qiu Mulan;Baohong(〔1.Kunming University of Science and Technology,Kunming,Yunnan 650093,China;Land and Resources Planning and Design Institute of Yunnan Province,Kunming,Yunnan 650216,China;Kunming Real Estate Ownership Investigation Center(Kunming Institute of Surveying and Mapping of Land Planning and Prospecting),Kunming,Yunnan 650200,China〕)

机构地区：[1]昆明理工大学国土资源工程学院,云南昆明650093 [2]云南省国土资源规划设计研究院,云南昆明650216 [3]昆明市不动产权籍调查中心(昆明市国土规划勘察测绘研究院),云南昆明650200

出　　处：《水土保持通报》2023年第1期141-148,共8页Bulletin of Soil and Water Conservation

基　　金：国家自然科学基金项目“滇中城市群国土格局多尺度演化模拟及优化配置”(41761081)。

摘　　要：[目的]对比分析逻辑回归(LogReg)、多层神经网络(MLP)和相似度加权学习(SimWeight)3种土地利用变化趋势模型在多地类变化分析中的模拟效果与土地利用预测精度,为云贵高原地区的国土空间规划、水土保持和生态修复等提供参考。[方法]以2000—2020年昆明市土地利用变化为例,分别采用3种模型对该区域建模,并使用受试者工作特征曲线、ROC曲线下面积和Kappa系数等多种方法评估精度。[结果]对于大多数地类的变化趋势,MLP和SimWeight模型的模拟效果要好于LogReg模型,特别是在未利用地的变化分析中MLP和SimWeight模型的AUC值均大于0.9。在整体土地利用变化预测上,LogReg,MLP和SimWeight3种模型的Kappa值分别为0.9066,0.9041,0.9253,整体预测结果表现接近,但SimWeight模型略微优于其他模型。[结论]对于昆明市LUCC建模的模型选择,若为追求更高精度可选择SimWeight模型,若更在意运算速度则优先选择MLP模型,若需要进一步分析驱动因子与土地变化的关系应选择LogReg模型。[Objective]The simulation effects and land use prediction accuracy of three land use transition models(LogReg,MLP,and SimWeight)for multi-land change analysis were compared and analyzed in order to provide a reference for land space planning,soil and water conservation,and ecological restoration in the Yunnan-Guizhou Plateau.[Methods]Three models were used to simulate land use changes for Kunming City from 2000 to 2020.Various methods such as ROC curve,AUC coefficient,and kappa coefficient were used to evaluate model performance.[Results]The simulations of change trends for most land use classes by MLP and SimWeight were better than the simulations by LogReg,especially with regard to the change analysis of unused land where the AUC values for MLP and SimWeight were greater than 0.9.The kappa values for LogReg,MLP,and SimWeight were 0.9066,0.9041,and 0.9253,respectively.The overall prediction results for the three models were similar,and SimWeight was slightly better than the two other models.[Conclusion]For LUCC modeling of Kunming City,SimWeight is recommended when the user desires higher accuracy,MLP is recommended when the user cares more about calculation speed,and LogReg is recommended when the user wants to further analyze the relationship between driving factors and land use changes.

关 键 词：土地利用变化 逻辑回归 多层神经网络 相似度加权学习 

分 类 号：F301.2[经济管理—产业经济]