标题: Exploring the potential of canopy reflectance spectra for estimating organic carbon content of aboveground vegetation in coastal wetlands
作者: Cheng, H (Cheng, Hang); Wang, J (Wang, Jing); Du, YK (Du, Yingkun); Zhai, TL (Zhai, Tianlin); Fang, Y (Fang, Ying); Li, ZH (Li, Zehui)
来源出版物: INTERNATIONAL JOURNAL OF REMOTE SENSING 卷: 42 期: 10 页: 3850-3872 DOI: 10.1080/01431161.2021.1883201 出版年: MAY 19 2021
摘要: Accurate estimation of organic carbon content or carbon storage of coastal wetland vegetation is essential for understanding the carbon cycle of coastal wetland ecosystems. This study aimed to explore the potential of canopy spectra of coastal wetland vegetation in estimating the organic carbon content of aboveground vegetation (OCCAV). A total of 54 representative vegetation quadrats were selected from coastal wetlands in Jiangsu Province, China, and their canopy reflectance spectra were measured using an in situ field spectrometer. In particular, two advanced spectral algorithms, fractional-order derivative (FOD) and optimal band combination algorithm, were used for spectral pre-treatments. Partial least squares regression (PLSR) and support vector machine (SVM) were employed to establish the OCCAV estimation models. Accuracies of the models constructed by the processed spectral parameters were compared with that of the models corresponding to five traditional pre-treatment methods and 15 common vegetation indices. Results showed that the FOD method captured subtler spectral characteristics than the first and second derivatives. The optimal estimation accuracies of PLSR and SVM models were obtained based on 0.75 and 1.50 order derivative spectra, respectively, and the ratios of performance to interquartile range (RPIQ) of the models were 2.57 and 2.97, respectively. Moreover, the optimal band combination algorithm effectively extracted the sensitive spectral parameters related to OCCAV, and the accuracy of the model established based on the spectral parameters extracted by this algorithm was generally better than that of the model obtained by the common vegetation indices. The optimal estimation result was achieved by the SVM model based on the optimized ratio vegetation index, with an RPIQ of 3.10. In summary, this research provides a theoretical basis for future studies on estimating the organic carbon and carbon storage of wetland vegetation based on large-scale canopy hyperspectral images and also helps to improve the knowledge of the carbon cycle of wetland ecosystems.
地址: [Cheng, Hang; Wang, Jing; Fang, Ying; Li, Zehui] Wuhan Univ, Sch Resource & Environm Sci, Wuhan, Peoples R China.
[Wang, Jing] Beijing Normal Univ, Coll Water Sci, Beijing 100875, Peoples R China.
[Du, Yingkun] Minist Agr & Rural Affairs Peoples Republ China, Acad Agr Planning & Engn, Beijing, Peoples R China.
[Zhai, Tianlin] Henan Agr Univ, Coll Resources & Environm Sci, Zhengzhou, Peoples R China.
通讯作者地址: Wang, J (通讯作者)，Beijing Normal Univ, Coll Water Sci, Beijing 100875, Peoples R China.
版权所有 © 武汉大学资源与环境科学学院