基于无人机高光谱遥感的淇澳岛红树林冠层叶片功能性状反演  

Inversion of Mangrove Canopy Leaf Functional Traits on the Qi'ao Island Based on UAV Hyperspectral Remote Sensing

作　　者：王萌 孙争争 何志东 王智慧 耿守保 赵新锋 杨龙 孙中宇 Wang Meng;Sun Zhengzheng;He Zhidong;Wang Zhihui;Geng Shoubao;Zhao Xinfeng;Yang Long;Sun Zhongyu(Zhuhai Ecological Environment Monitoring Station of Guangdong Province,Zhuhai 519070,China;Qiao-DanGan Island Provincial Nature Reserve Management Office,Zhuhai 519000,China;Guangdong Open Laboratory of Geospatial Information Technology and Application//Guangzhou Institute of Geography,Guangdong Academy of Sciences,Guangzhou 510070,China)

机构地区：[1]广东省珠海生态环境监测站,广东珠海519070 [2]广东珠海淇澳-担杆岛省级自然保护区管理处,广东珠海519000 [3]广东省地理空间信息技术与应用公共实验室//广东省科学院广州地理研究所,广州510070

出　　处：《热带地理》2023年第11期2146-2154,共9页Tropical Geography

基　　金：广东省科学院建设国内一流研究机构行动专项资金项目(2020GDASYL-20200102001,2020GDASYL-0104002);广东省林业改革发展资金项目。

摘　　要：以珠海淇澳岛红树林为研究对象,基于无人机高光谱数据,采用偏最小二乘回归方法和归一化植被指数相结合(PLSR+NDVI)以及偏最小二乘回归方法与连续小波变换(PLSR+CWT)相结合的2种无人机高光谱数据处理方法,反演了研究区内红树林的10种冠层叶片功能性状。结果表明,PLSR+NDVI的方法更适用于红树林冠层叶片比叶重LMA、单位质量磷含量P_(mass)和单位面积氮含量N_(area)的反演,PLSR+CWT的方法更适用于氮磷比N/P、叶绿素含量C_(ab)和类胡萝卜素含量C_(xc)的反演,2种方法用于反演单位质量氮含量N_(mass)、单位质量钾含量K_(mass)、单位面积磷含量P_(area)和单位面积钾含量K_(area)的结果均不理想(R~2<0.3)。采用所建立的较优方法对研究区红树林冠层叶片的LMA、P_(mass)、N_(area)、N/P、C_(ab)和C_(xc)的含量进行反演和空间分布制图,得到的功能性状空间分布格局与冠层结构及物种的空间分布格局密切相关。将不同生活型的红树林物种分开构建高光谱功能性状反演模型,有望进一步提升模型的反演精度。未来可针对每一个红树林物种构建特定的功能性状反演模型,同时结合可见光影像的物种识别结果,以提升红树林冠层叶片功能性状的反演精度。Quantitative studies of mangrove leaf functional traits will help us understand the adaptive evolutionary strategies of mangrove plants and the relationship between mangrove biodiversity and ecosystem functions.Because of the special environment of the intertidal zone where mangroves are located,it is very difficult to obtain the functional traits of mangrove canopies from the ground,and relevant studies are lacking.The maturity of Unmanned Aerial Vehicle(UAV)hyperspectral remote sensing technology provides a new means of conducting such research.This study considered mangroves on Qi'ao Island,Zhuhai,as the research object.Based on UAV hyperspectral data,two UAV hyperspectral data processing methods,which combined Partial Least Squares Regression with Normalized Difference Vegetation Index(PLSR+NDVI)and Partial Least Squares Regression with Continuous Wavelet Transform(PLSR+CWT),were used to estimate the 10 canopy leaf functional traits of mangroves on Qi'ao Island.The results showed that the PLSR+NDVI method was more suitable for the inversion of mangrove canopy-specific leaf weight(LMA),phosphorus content per unit mass(Pmass),and nitrogen content per unit area(N_(area)),whereas the PLSR+CWT method was more suitable for the estimation of the nitrogen/phosphorus ratio(N/P),chlorophyll content(C_(ab)),and carotenoid content(C_(xc)).However,the results of the above two methods for retrieving the nitrogen content per unit mass(N_(mass)),potassium content per unit mass(K_(mass)),phosphorus content per unit area(Parea),and potassium content per unit area(K_(area))were not ideal(R^(2)<0.3).The optimal method established in this study was used to estimate the contents of LMA,P_(mass),N_(area),N/P,C_(ab),and C_(xc) of the mangrove canopy leaves in the study area and map their spatial distribution.Mangrove canopy leaf functional traits obtained using UAV hyperspectral data inversion better reflect the horizontal structure and function of the mangrove community.Regarding the spatial distribution patterns of canopy leaf fun

关 键 词：叶片功能性状 无人机高光谱遥感 功能性状制图 PLSR 红树林 淇澳岛 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置] S451[自动化与计算机技术—控制科学与工程]