基于光谱特征的雅鲁藏布大峡谷森林枯落物水源涵养能力反演研究  

Research on Inversion of Water Conservation Capacity of Forest Litter in Yarlung Zangbo Grand Canyon Based on Spectral Features

作　　者：龙芊芊 周仁浩 岳德鹏[1] 牛腾 毛雪晴 王朋冲 于强[1] LONG Qian-qian;ZHOU Ren-hao;YUE De-peng;NIU Teng;MAO Xue-qing;WANG Peng-chong;YU Qiang(Beijing Key Research Office of Precision Forestry,Beijing Forestry University,Beijing 100083,China;Cyberspace Security Academy,Chengdu University of Information Technology,Chengdu 610200,China;Beijing Linmiao Ecological Environment Technology Co.,Ltd.,Beijing 100085,China)

机构地区：[1]北京林业大学精准林业北京重点研究室,北京100083 [2]成都信息工程大学网络空间安全学院,四川成都610200 [3]北京林淼生态环境技术有限公司,北京100085

出　　处：《光谱学与光谱分析》2022年第1期229-235,共7页Spectroscopy and Spectral Analysis

基　　金：西藏林芝色季拉山森林生态系统水源涵养服务功能评估及辐射效应研究项目(XZA-JYBSYS-2019-01)资助。

摘　　要：水源涵养是生态系统重要的服务功能,森林作为一种复杂的生态系统,其组成部分对于水源涵养的贡献率各不相同;森林枯落物直接覆盖于地表,既来自于林冠层也抑制土壤层的水分蒸散,因此枯落物层在水源涵养功能中发挥了重要作用。遥感和高光谱技术为远距离识别面状区域的水源涵养能力提供了解决方案,特别在高原地区,遥感是获取地表信息的快速手段。以雅鲁藏布大峡谷为研究区,用ASD(便携式地物)光谱仪测定主要树种(高山松、林芝云杉和川滇高山栎)的叶片高光谱数据并构建植被指数,同时,通过样地采样获取枯落物样本并计算样本的持水拦蓄性能,然后建立植被指数与有效拦蓄量的多元回归模型。在此基础上,基于Sentinel-2影像反演大峡谷主要树种枯落物的水源涵养能力分布情况,最后结合验证点对反演模型进行精度评价。结果显示:(1)三类树种的叶片反射率趋势相似,川滇高山栎的反射率最高,高山松次之,林芝云杉最低;(2)枯落物的有效拦蓄量从大到小排序为:林芝云杉(48.36 t·ha^(-1))>川滇高山栎(39.24 t·ha^(-1))>高山松(32.32 t·ha^(-1))。林芝云杉枯落物的分解程度和蓄积量均最高,因此持水拦蓄能力最强;川滇高山栎的革质叶片不利于分解堆积,进而限制蓄水能力;高山松含有较多油脂,不易被水浸湿,导致持水能力较弱。(3)通过Person相关系数分析和多元线性回归模型得知,叶片蜡质参数和衰减程度越高,枯落物的水源涵养能力越弱;植被生长态势越好、色素和叶片水分含量越高,其水源涵养能力越强。(4)枯落物水源涵养能力反演模型的精度评价结果良好,高山松、林芝云杉和川滇高山栎的样本检验点拟合优度R^(2)分别为0.943,0.815和0.812,均方根误差RMSE分别为1.597,2.270和1.953,表明模型可以用于大峡谷森林枯落物水源涵养能力的预测分布研究。Water conservation is an important part of the ecosystem service function.As a complex ecosystem,forests have different contributions to water conservation.The forest litter plays an important role in the water conservation function,for it directly covers the ground surface.Remote sensing and hyperspectral technology provide a solution for the long-distance detection of water conservation in planar areas.Especially in plateau areas,remote sensing is the most effective way to obtain surface information.In this paper,the hyperspectral data of main tree species were measured by ASD spectrometer in the Grand Canyon research area of Yarlung Zangbo.The litter samples were obtained by sampling on the plot and the water holding capacity of the samples was calculated.Use leaf spectrum information to construct vegetation index related to litter water conservation capacity,then establish multiple regression model of vegetation index and effective retention capacity,and invert the water conservation capacity distribution of main tree species in the Grand Canyon based on Sentinel-2 images.In the end,the accuracy of the inversion model is evaluated based on the verification points.The results showed that:(1)The reflectance of Quercusaqui folioides was the highest,the lowest was the Picea Linzhi,and the total reflectance trend of the three species was similar;(2)The effective interception amount of litter is sorted from large to small as follows:Picea Linzhi(48.36 t·ha^(-1))>Quercusaqui folioides(39.24 t·ha^(-1))>Pinus densata(32.32 t·ha^(-1)).Picea Linzhi leaves are easy to decompose and store,Quercusaqui folioides’leathery leaves and Pinus densata’s oily leaves are not conducive to decomposition.Therefore,the litter of Picea Linzhi has the highest effective interception amount.(3)Through the Person correlation coefficient analysis and the multiple linear regression model,it is found that the higher the wax parameters and attenuation degree of leaves,the weaker the water conservation capacity of litter;the better the ve

关 键 词：高光谱 水源涵养 枯落物 遥感反演 雅鲁藏布大峡谷 

分 类 号：S715[农业科学—林学]