Spectral Characteristics And Retrieval Of Photosynthetic Pigments In Wetland Vegetation Leaves

The wetland ecosystem is an important part of the terrestrial ecosystem,and plays an important role in the material circulation of carbon and nitrogen energy exchange.Photosynthetic pigments such as chlorophyll and carotenoids contained in the leaves of wetland plants are important biochemical parameters affecting the photosynthetic process.The spectral characteristics they exhibit on the leaves provide a possibility for rapid and non-destructive estimation of their contents.It is one of the forefront research directions for quantitative remote sensing of vegetation.However,the current research on quantitative inversion of photosynthetic pigments based on leaf spectra mainly focuses on farmland and forests.There are few studies on wetland ecosystems,especially the lack of studies on the inversion of carotenoids.In this study,Phragmites australis?local species,C3 plants?and Spartina alterniflora?foreign species,C4 plants?,typical species of Dongtan wetland in Shanghai,were used as examples.Using the data from the actual measurement and model simulation of the leaves,the chlorophyll and carotenoid content were inversely studied.The main research contents and results of this paper were as follows:1.Taking Chongming Dongtan Wetland as the research area,the field experiment for Phragmites australis and Spartina alterniflora was designed and carried out.The contents of chlorophyll,carotenoids and the spectral reflectance?near-infrared and visible bands?of leaves of the two planted leaves were measured,and the corresponding measured database was established.2.In the spectral data processing,the spectral reflectance data of the original leaf was decomposed and reconstructed using Empirical Mode Decomposition?EMD?.The results showed that the reflectance data and the differential reflectance data after decomposition and reconstruction had higher signal-to-noise ratio,particularly noticeable in the visible bands such as blue light and red light.On this basis,the full-band hyperspectral vegetation indices RVI?_1,?₂ and NDVI?_1,?₂,the trilateral parameters?red,yellow,and blue?,the morphological parameters of the red and green peaks,the relative reflectance after the continuum removal and the Angular Vegetation Index?AVI?were calculated.The multivariate stepwise regression method?Linear method?and BP neural network method?Nonlinear methods?were used to study the chlorophyll content and carotenoid content,respectively.3.For chlorophyll inversion,three methods of univariate linear regression,multiple stepwise linear regression and BP neural network were adopted.The results showed that:For Phragmites australis,BP neural network method had the highest precision,R² and RMSE were 0.80 and 2.74 respectively.The univariate model based on the whole-band vegetation index was the second,R² and RMSE were 0.76 and2.89 respectively.The R² and RMSE of the multivariate stepwise linear regression were 0.69 and 4.68,respectively,but the principal component analysis was used for preprocessing,and the first and second principal components were selected into the model.The inversion accuracy was significantly improved,and R² was increased by4.3%and RMSE reduced by 29.7%.The chlorophyll inversion results of Spartina alterniflora also presented similar results.4.For the inversion of carotenoids,the AVI index constructed with different wavelength combinations was superior to other indexes in the retrieval of leaf carotenoids in terms of index selection.In terms of method comparison,the BP neural network method had the highest accuracy,R² and RMSE of the Phragmites australis carotenoid inversion were 0.85 and 0.45,respectively,which was superior to the multiple regression methods of 0.67 and 0.57.After the principal component transformation,the inversion accuracy of the linear regression model established using the first principal component as an independent variable was better than the multiple regression model,R² and RMSE were 0.76 and 0.53,respectively.Similar results were obtained for inversion of carotenoids from Spartina alterniflora based on different methods.It can be seen that the principal component analysis was similar to the above chlorophyll inversion in improving the inversion accuracy of carotenoids,and had also played a positive role.