| Chlorophyll a content is one of the important indicators of water environment assessment,which has important significance for the primary productivity and eutrophication of water bodies.Quantitative inversion of chla in inland water bodies has always been a hot issue in related fields.Unlike ocean water bodies,inland water bodies have complex composition due to the input of large amounts of terrestrial materials,especially suspended matter（ss）,colored soluble organic matter（colored soluble organic matter）.Soluble organic matter（CDOM）has obvious interference on the research of chla inversion.Therefore,how to accurately invert chla for inland water bodies is one of the important research directions in the field of aqua remote sensing.Aiming at the above problems,this dissertation proposes a chla inversion method based on multivariate curve resolution and Lorentz spectrum shape constraints.In October 2020,a black barrel experiment was carried out in Huahang Lake,with different concentration levels of chla and ss water samples,the system obtained the measured spectrum and water quality parameter data of the water samples,and the spectral data was processed and simulated to obtain the Landsat 8 band value.Based on the linear decomposition model of the mixed spectrum,and assuming that the functional relationship between ss and chla and spectral reflectance is a linear relationship,through correlation analysis and water quality parameter spectral feature analysis,the best sensitive band combination of chla is determined,based on the above assumptions and data processing On the basis of,the Alternating Least Squares Algorithm（MCR-LALS）with multivariate curve resolution and Lorentz spectral shape constraint is used to realize the quantitative inversion of chla from remote sensing.The experimental results show:（1）B3,B2/B1,B3/B1,B3/B2,B3/（B1+B4）have a relatively obvious linear relationship to ss and chla at the same time.After baseline deduction,the above bands can meet the convergence conditions of the MCR-LALS model;（2）The MCR-LALS model can decompose the input band information into the characteristic spectrum of water quality parameters and a coefficient matrix.The coefficient matrix contains water quality parameter concentration information and can be used for quantitative inversion of chla;（3）Using data analysis software to linearly fit the chla coefficient matrix with the measured concentration of chla,the fitting effect is better,the fitting coefficient is 0.798,and the root mean square error is 2.398mg·L-1;（4）The model accuracy test is performed with the verification sample,which is compared with the traditional two Compared with the band model（B3/B1）and the three-band model（B3/（B1+B4））,the method proposed in this paper has the highest accuracy,indicating that the method has better practicability,especially for inland turbid water bodies.Based on the idea of mixture spectrum decomposition,starting from the principle of MCR-LALS,the alternate least squares method of multivariate curve resolution and Lorentz spectrum shape constraint is used to decompose the water separation reflectance to obtain the chlorophyll a spectrum curve and its coefficient matrix.Using data analysis software,a linear regression model was established based on the decomposed chla coefficient matrix and the measured chla concentration.According to the fitting coefficient（R2）,the average relative error（ARE）and the root mean square error（RMSE）,the two-band algorithm,the three-band algorithm and the MCR-LALS algorithm are analyzed and compared,and the results show that the MCR-LALS algorithm has higher inversion accuracy.This dissertation explores the inversion strategy of chlorophyll a from the perspective of spectral decomposition,establishes a chla inversion method based on the MCR-LALS model,and achieves good experimental results.This method is a brand-new exploration and attempt,which will provide new ideas for the inversion of chlorophyll a in turbid inland waters in the future. |
PDF Full Text Request