基于地理加权随机森林的青藏地区放牧强度时空格局模拟  

Mapping the multi-temporal grazing intensity on the Qinghai-Tibet Plateau using geographically weighted random forest

作　　者：李兰晖 黄聪聪 张镱锂[2] 刘林山[2] 王兆锋[2] 张海燕[2] 丁明军[3] 张华敏 Li Lanhui;Huang Congcong;Zhang Yili;Liu Linshan;Wang Zhaofeng;Zhang Haiyan;Ding Mingjun;Zhang Huamin(School of Computer and Information Engineering,Xiamen University of Technology,Xiamen 361024,Fujian,China;Key Laboratory of Land Surface Pattern and Simulation,Institute of Geographic Sciences and Natural Resources Research,Chinese Academy of Sciences,Beijing 100101,China;School of Geography and Environment,Jiangxi Normal University,Nanchang 330022,Jiangxi,China)

机构地区：[1]厦门理工学院计算机与信息工程学院,福建厦门361024 [2]中国科学院地理科学与资源研究所陆表格局与模拟重点实验室,北京100101 [3]江西师范大学地理与环境学院,江西南昌330022

出　　处：《地理科学》2023年第3期398-410,共13页Scientia Geographica Sinica

基　　金：中国科学院战略性先导科技专项(XDA20040201);第二次青藏高原综合科学考察研究(2019QZKK0603);国家自然科学基金项目(42101099)资助。

摘　　要：精确刻画放牧强度时空格局是深刻理解青藏地区高寒草地服务功能动态及其驱动机制的重要基础,对辅助制定区域生态安全和可持续发展战略具有意义。基于牲畜年末存栏量、牧区人口密度、生长季NDVI、年降水量、年平均气温和居民点分布等数据,采用地理加权随机森林模型,模拟了2000年、2010年和2020年3期青藏地区放牧强度的空间格局,并定量评价了环境因子对放牧强度空间分布解释性的区域差异。结果表明:(1)地理加权随机森林模拟结果可精细地刻画青藏地区放牧强度的空间特征,与经典随机森林模型相比,判定系数更高,平均绝对值误差和均方根误差更低。(2)青藏地区放牧强度呈现东南高、西北低的基本特征,其中,西北部地区放牧强度低于25羊单位/km~2的区域约占青藏地区面积的1/2。(3)与2000年和2010年相比,2020年青藏地区放牧强度呈现总体下降、局部抬升的态势;其中,较2010年,2020年放牧强度下降超过1羊单位/km~2的区域占牧区面积的61.69%。(4)牧区人口密度是解释放牧强度空间异质性最主要的因素,其相对重要性呈现西高东低的特征,而降水量和生长季NDVI的相对重要性则呈现西北高、东南低的特征。研究结论可为青藏地区草地可持续管理和生态安全屏障建设提供科学参考。Accurately quantifying the spatiotemporal pattern of grazing intensity on the Qinghai-Tibet Plateau(QTP)is crucial to improving our understanding of the driving mechanism of alpine grassland change,and is of great significance for maintaining regional ecological security and promoting sustainable development policies.Based on the data of livestock inventory at the end of the year and environmental covariates(e.g.population density in pastoral areas,growing season NDVI,annual precipitation,annual mean temperature,and settlements),the gridded grazing intensities on the QTP in 2000,2010,and 2020 were simulated by geographically weighted random forest(GRF),and the regional differences in the interpretability of environmental variables were then analyzed.The results showed that grazing intensity maps predicted by the GRF could mirror the spatial distribution of grazing intensity on the QTP,compared with the classic random forest model,the R2 was higher,and both the mean absolute value error(MAE)and root mean square error(RMSE)were lower.The grazing intensity was generally higher in the southeast and lower in the northwest of the QTP.The areas with grazing intensity of less than 25 sheep units/km^(2) in the northwest part accounted for about half of the QTP.Compared with 2000 and 2010,the grazing intensity of the QTP in 2020 showed a trend of overall decrease but local increase.For example,compared with 2010,the areas of grazing intensity decreased by higher than 1 sheep unit/km^(2) in 2020 accounting for 61.69%of the pastoral area.The population density in pastoral areas was the most important factor explaining the spatial heterogeneity of grazing intensity,and its relative importance was higher in the western QTP and lower in the eastern QTP.On the contrary,the relative importance of both precipitation and growing season NDVI was higher in the northwest part and lower in the southeast part of QTP.Our results provide scientific references for sustainable grassland management and ecological safety barrier construction

关 键 词：放牧强度 地理加权随机森林 牧区人口密度 青藏地区 

分 类 号：S17[农业科学—农业基础科学]