A novel method for leaf chlorophyll retrieval based on harmonic analysis: a case study on Spartina alterniflora

Spartina alterniflora is the main invasive vegetation in wetland ecosystems, and information on its chlorophyll content is important data for quantitative research on the key ecological functions of wetland ecosystems. The Dongtan wetland of the Yangtze River estuary was used as the experimental research area, and the artificial cultivation field of Eupatorium was taken as the research object. To prevent uncertainty in the calculation of the chlorophyll content inversion factor for Spartina alterniflora leaves, harmonic analysis theory was adopted in this study. First, the original measured spectral data in the range of 400 nm ~ 1000 nm were decomposed and reconstructed by empirical mode decomposition (EMD), and the harmonic characteristic parameters of the EMD reconstruction spectral data were obtained by harmonic analysis (HA). Meanwhile, using the PROSPECT-D radiation transfer model, the simulated data was used to validate the inversion model. Based on these harmonic characteristic parameters, harmonic analysis-back propagation (HA-BP) and stepwise multiple linear regression (SMLR) models were established. Finally, the model was validated with data simulated by the PROSPECT-D model and compared the measured chlorophyll contents using the inversion values of the two models. The results show that the inversion accuracy of the HA-BP model was highest for the measured data. The determination coefficient (R2) and root mean square error (RMSE) of the model were 0.8528 and 6.8968 (μg/cm2), respectively, and the inversion accuracy of the simulated data was slightly lower than that of the measured data. The results show that the original spectral noise could be effectively suppressed by EMD and reconstruction, while harmonic analysis could compress the signal and prevent uncertainty in the spectral parameter calculation. Thus, the harmonic decomposition method could be applied to the inversion of Spartina alterniflora chlorophyll contents. The methods in this research provide a theoretical basis and technical support for the use of frequency domain parameters for the inversion of vegetation pigment contents.