Tropical Geography


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

Meng Wang1(), Zhengzheng Sun2, Zhidong He1, Zhihui Wang3, Shoubao Geng3, Xinfeng Zhao1, Long Yang3, Zhongyu Sun3()   

  1. 1.Zhuhai Ecological Environment Monitoring Station of Guangdong Province, Zhuhai 519070, China
    2.Qiao-DanGan Island Provincial Nature Reserve Management Office, Zhuhai 519000, China
    3.Guangdong Open Laboratory of Geospatial Information Technology and Application//Guangzhou Institute of Geography, Guangdong Academy of Sciences, Guangzhou 510070, China
  • Received:2023-07-27 Revised:2023-09-05 Online:2023-10-31
  • Contact: Zhongyu Sun;


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 (Narea), whereas the PLSR + CWT method was more suitable for the estimation of the nitrogen/phosphorus ratio (N/P), chlorophyll content (Cab), and carotenoid content (Cxc). However, the results of the above two methods for retrieving the nitrogen content per unit mass (Nmass), potassium content per unit mass (Kmass), phosphorus content per unit area (Parea),and potassium content per unit area (Karea) were not ideal (R2<0.3). The optimal method established in this study was used to estimate the contents of LMA, Pmass, Narea, N/P, Cab, and Cxc 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 functional traits, the spatial distribution patterns of Narea, Cab, Cxc, and N/Pwere relatively consistent with higher values in the middle region and lower values in the edge region. The spatial distribution patterns of the LMA and Pmass were similar, and the distribution was relatively uniform throughout the study area. Combined with ground survey data, the internal relationship between species composition and spatial patterns of functional traits, as well as ecosystem functions and processes, can be deeply explored, and rapid investigation and assessment of mangrove forests can be realized at the community and ecosystem scales. The spatial distribution pattern of functional traits was closely related to the spatial distribution pattern of canopy structure and species. The inversion model of hyperspectral functional traits was constructed by separating mangrove species with different life types, which is expected to further improve the inversion accuracy of the model. Constructing a specific functional trait inversion model for each mangrove species, combined with the species identification results of visible-light images, will effectively improve the inversion accuracy of mangrove canopy leaf functional traits.

Key words: leaf functional traits, UAV hyperspectral remote sensing, functional trait mapping, PLSR, mangrove forest, Qi'ao Island

CLC Number: 

  • TP751