| Autor: |
Yuchao, Fu, Wanxiang, Li, Tianyuan, Liu, Zhen, Zhang, Haochen, Li, Jingran, Xu, Meizhen, Huang |
| Rok vydání: |
2023 |
| Předmět: |
|
| Zdroj: |
Science of The Total Environment. 865:160950 |
| ISSN: |
0048-9697 |
| Popis: |
Fluorescent substances exist in various aquatic environments and other environmental media. It is a critical task to identify the components accurately and quantify their contents precisely. Based on the Crosstalk Fluorescence Spectroscopy Analysis (CFSA) model, a fluorescence spectroscopic decomposition using the Alternating Gradient Descent (AGD) algorithm is developed. By reducing the residual error of the model through alternating iterations, the CFSA-AGD method achieves unsupervised model training and automatic spectroscopic decomposition without extra experimental operations such as dilution or absorbance measurement, exempting from tedious modeling process. The objectives of this work are to validate that the CFSA-AGD method can comprehensively address the decomposition of fluorescence spectral crosstalk. Furthermore, the novel method is applied to the spectroscopic decomposition of natural FDOMs in aquatic environments as a standard tool. The spectral data analyzing the performance of this method is verified and compared with the conventional methods through the experiment on standard samples. The results indicate that CFSA-AGD has higher spectroscopic decomposition accuracy and gives more abundant information on the characteristic spectra with less residual error than parallel factor analysis. This means that the fluorescence spectra of natural FDOMs can be decomposed into the characteristic fluorescence emission spectra of single components with higher accuracy and the characteristic fluorescence absorption spectra that cannot be obtained by the conventional methods. Meanwhile, it improves the analytical precision of the contents (from R |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect