基于无人机多光谱遥感的矮林芳樟光合参数估测  

Estimation of Photosynthetic Parameters of Cinnamomum camphora in Dwarf Forest Based on UAV Multi-spectral Remote Sensing

作　　者：鲁向晖[1] 龚荣新 张海娜 王倩 张杰[1] 谢荣秀 LU Xianghui;GONG Rongxin;ZHANG Haina;WANG Qian;ZHANG Jie;XIE Rongxiu(Jiangxi Provincial Engineering Research Center of Seed-breeding and Utilization of Camphor Trees,Nanchang Institute of Technology,Nanchang 330099,China;Jiangxi Provincial Technology Innovation Center for Ecological Water Engineering in Poyang Lake Basin,Nanchang 330029,China)

机构地区：[1]南昌工程学院江西省樟树繁育与开发利用工程研究中心,南昌330099 [2]江西省鄱阳湖流域生态水利技术创新中心,南昌330029

出　　处：《农业机械学报》2023年第10期179-187,共9页Transactions of the Chinese Society for Agricultural Machinery

基　　金：国家自然科学基金项目(52269013、32060333);江西省教育厅科技项目(GJJ190977、GJJ190946);江西省主要学科学术和技术带头人培养计划青年项目(20204BCJL23046);江西省科技厅重大科技专项(20203ABC28W016-01-04);江西省林业局樟树研究专项(202007-01-04)。

摘　　要：为探讨应用无人机多光谱技术估算矮林芳樟(Cinnamomum camphora(Linn.)Presl)光合参数的有效分析模型和方法,本研究以矮林芳樟为研究对象,通过无人机搭载的多光谱相机获取其冠层六波段光谱反射率,同步测量其净光合速率(Pn)、胞间二氧化碳浓度(Ci)、气孔导度(Gs)和蒸腾速率(Tr)4种光合参数,采用最佳指数因子(OIF)筛选光谱反射率和植被指数的组合作为自变量,分别采用偏最小二乘法(Partial least squares method,PLS)、反向传播神经网络(Back propagation neural network,BPNN)和随机森林(Random forest,RF)构建自变量与光合参数的估算模型,并分析比较各估算模型的精度。结果显示:矮林芳樟光合参数与叶片红边波段2(中心波长750 nm)和近红外波段(中心波长840 nm)反射率有密切关系;红边波段2、增强型植被指数2(EVI2)、红边叶绿素指数(CIrededge)组合的OIF值最大,为0.0126,可作为模型自变量的最佳组合;Pn、Ci、Gs、Tr 4种光合参数的最优模型均为BPNN,其建模集决定系数R^(2)分别为0.85、0.81、0.80、0.82,均方根误差(RMSE)分别为0.85μmol/(m^(2)·s)、16.23μmol/mol、0.03 mol/(m^(2)·s)、0.37 mmol/(m^(2)·s),相对分析误差(RPD)分别为2.59、2.33、2.28、2.37;验证集R^(2)为0.81、0.73、0.83、0.76,RMSE为1.46μmol/(m^(2)·s)、18.37μmol/mol、0.03 mol/(m^(2)·s)、0.67 mmol/(m^(2)·s),RPD为1.39、1.86、2.67、1.20。研究结果可为无人机多光谱遥感矮林芳樟光合参数估测提供理论依据,为快速监测大面积经济植物生长状况提供技术支撑。In order to explore an effective analytical model and method for estimating photosynthetic parameters of Cinnamomum camphora(Linn.)Presl by using unmanned aerial vehicle(UAV)multispectral technology,taking Cinnamomum camphora(Linn.)Presl as the research object,its canopy six-band spectral reflectance was obtained through a multispectral camera carried by UAV,and its net photosynthetic rate(Pn),intercellular carbon dioxide concentration(Ci),stomatal conductance(Gs)and transpiration rate(Tr)were simultaneously measured.The optimal index factor(OIF)was used to screen the combination of spectral reflectance and vegetation index as independent variables.Partial least squares method(PLS),back propagation neural network(BPNN),and random forest(RF)were used to construct estimation models for the independent variables and photosynthetic parameters,and the accuracy of each estimation model was analyzed and compared.The results showed that there was a close relationship between photosynthetic parameters and leaf reflectance in the red edge band 2(center wavelength 750nm)and near infrared band(center wavelength 840nm)of Cinnamomum camphora L.The combination of red edge band 2,enhanced vegetation index 2(EVI2),and red edge chlorophyll index(CI rededge)had the highest OIF value of 0.0126,which can be used as the best combination of model independent variables.The optimal models for the four photosynthetic parameters Pn,Ci,Gs,and Tr were all BPNN,with the modeling set decision factors R^(2)of 0.85,0.81,0.80,and 0.82,and the root mean square error(RMSE)of 0.85μmol/(m^(2)·s),16.23μmol/mol,0.03mol/(m^(2)·s)and 0.37mmol/(m^(2)·s).The relative analytical error(RPD)were 2.59,2.33,2.28,and 2.37,respectively.The R^(2)of the validation set was 0.81,0.73,0.83,0.76,and the RMSE was 1.46μmol/(m^(2)·s),18.37μmol/mol,0.03mol/(m^(2)·s)and 0.67mmol/(m^(2)·s),with RPD of 1.39,1.86,2.67,and 1.20,respectively.The research results can provide a theoretical basis for the estimation of photosynthetic parameters of Cinnamomum camphora in dw

关 键 词：矮林芳樟 光合参数 估算 多光谱 植被指数 无人机遥感 

分 类 号：S758[农业科学—森林经理学]