Dissertation > Excellent graduate degree dissertation topics show

Fluorescence Discrimination Technique Based on Wavelet Analysis for the Phytoplankton in the Coastal Waters of China

Author: LiuBao
Tutor: SuRongGuo
School: Ocean University of China
Course: Analytical Chemistry
Keywords: phytoplankton fluorescence spectra wavelet analysis bayes analysis nonnegative least squares
CLC: X834
Type: Master's thesis
Year: 2009
Downloads: 17
Quote: 1
Read: Download Dissertation

Abstract

Phytoplankton are microscopic plants which are aquatic ,unicellular. Some of them are the reason of harmful algal blooms (HABs). Recently, the HABs had become a very severe problem in the coastal waters of china, and retrick the develop of the littoral’s economy. It had became an urgent need for determining the algae causing HABs in actual locale, rapidly and reliably, so that we can take avail measure to reducing the influence of HABs. The excitation-emission matrixes (EEMs) can provide abundant information ,at the same time have high sensitivity、selectivity and needless of complicated pre-dispose. It can satisfy the requirements of actual locale and real time.Based on the analysis of the characteristics of the phytoplankton communities in coastal waters of china, in this paper forty-three algae which are familiar in coastal waters of china are selected and cultured in the lab. They belong to seven divisions and thirty-two genera .After measuring the 3D-fluoresscence spectra, analysis the stability of the EEMs of the same phytoplankton and the difference between EEMs of different phytoplankton, then expound the feasibility of utilization the EEMs in the identification of the familiar phytoplankton of Chinese coastal water. Based on the feasibility, three wavelets are selected to be utilized in the analysis of the EEMs of phytoplankton. The identifying methods of the familiar phytoplankton species of Chinese coastal water are established with utilization of wavelet analysis, At the same time established the primary semi-quantitative fluorescence method. The main conclusions of this paper are as follows:1. Utilizing the Relative Standard Deviation (R.S.D.) to analysis the stability of EEMs of the same phytoplankton, and the Analysis of Variance (ANOVA) to analysis the difference of the EEMs of different phytoplankton. The results present that most phytoplankton present good stabilities in the same species and significance difference between different species. The method based on the EEMS of phytoplankton to identifying the species is feasible.2. Two orthogonal wavelets (ciof2 and symlet7) which have better ability of feature extraction are selected to decompose the EEMs of the phytoplankton. Clustering analysis is utilized to extract the reference spectra to establish two-rank database of reference spectra on the level of division and genera. Based on the databases, a fluorescence method for identification of phytoplankton species is established with utilizing the Nonnegative Least Squares (NNLS). The results the method can identifying :more than 92% at the level of division and more than 83% at the level of genus. Some phytoplankton species can be identified at the level of genus by the fluorescence method when the cell density above 5x106 cell L-1 in actual sea waters samples.3. Bior1.1 which has better ability of feature extraction are selected from the bi-orthogonal wavelets. Utilizing the Clustering Analysis and Nonnegative Least Squares (NNLS) to establish fluorescence method for identification of phytoplankton. The results present that the identification of the identifying ability of bior1.1 is not as good as the orthogonal wavelets on the whole, but have better identifying ability at some phytoplankton which the orthogonal wavelets could not get satisfactory results. The bior1.1 wavelet is complementary of the orthogonal wavelet.4. Comparing the three identifying technology based on different wavelets, could get the conclusion that the coif2 wavelet have better ability of identification than other two wavelets, But the symlet7 and the bior1.1 wavelets could get better result at some phytoplankton species which the coif2 wavelet could not give satisfying results. Based on the complementary of three wavelets, we can get : more than 95% at the level of division and more than 85% at the level of genus.The innovation of this paper is the establishment of the two-rank reference spectra for the main phytoplankton on the level of division and genus based on two orthogonal wavelets and one bi-orthogonal wavelet, really achieved the joint use of the two-rank reference spectra databases. Based on the complementary of three wavelets, establish multi-rank reference spectra databases.

Related Dissertations

CLC: > Environmental science, safety science > Environmental Quality Assessment and Environmental Monitoring > Environmental monitoring > Marine monitoring
© 2012 www.DissertationTopic.Net Mobile