Estimation of above-ground biomass of reed (Phragmites communis) based on in situ hyperspectral data in Beijing Hanshiqiao Wetland, China

• DOI: 10.1007/s11273-018-9644-5
• Publication Date: 2018-11-20T09:51:59Z
• Authors: Wei Li; Zhiguo Dou; Yan Wang; Gaojie Wu; Manyin Zhang; Yinru Lei; Yunmei Ping; Jiachen Wang; Lijuan Cui; Wu Ma
• Keywords: Phragmites; Swamp; Empirical modelling

Accurate estimates of reed (Phragmites communis) biomass are critical for efficient reed swamp monitoring and management. This study compared the accuracy of commonly used empirical models in estimating above-ground biomass in dense swamp reeds in the Beijing Hanshiqiao Wetland Nature Reserve, northern China. Two-thirds of the samples were used for model construction, and one-third for model validation. Models for estimating reed above-ground biomass, based on original spectral reflectance, first-order differential spectrum, trilateral parameters and partial least squares (PLS), were constructed using univariate linear regression and the PLS method. Results showed that the biomass estimation model based on the first-order differential spectrum was relatively inefficient. Model accuracy was highest in the PLS model, followed by the original spectral reflectance model and was lowest in the trilateral parameters model. The model validation results were consistent with the accuracy of the established estimation model, so the model has good stability. We conclude that above-ground biomass can be successfully estimated using canopy hyperspectral information on wetland plants, based on the empirical model. The PLS method not only was more accurate in estimating fresh biomass but also represented a significant improvement in estimating dry biomass.