Studies On Characteristics Of Absorption Spectra For Identifying Phytoplankton Classes

With the frequent occurances of red-tide in coastal water of our country at present.studies on monitoring phytoplankton classes and their abundance have already become the focus. Being cheap,steady,and easy to be handled.UV-Spectrophotometer has been applied in enviromental analysis extensivly.It is very significant to determine red-tide of the East China Sea fast and effectivly by using Absorption Spectrum method. This paper aims to study the characteristics of the absorption spectra of phytoplankton. Nine phytoplankton species, representing five divisions of the typical or dominant red-tide species of the East China Sea, were chosen and grown in lab cultures under different temperatures(20℃, 15℃) and different illumination intensities (7000lux, 1100lux) . The fingerprint features are extracted to identify these different phytoplankton species . The main research work is as follows:1. There is influence on the absorption spectrum of the nine kinds of algae under different temperature and illumination intensity: During different periods, the spectral shape of Skeletonema costatuma and Synechococcus sp will changed greatly.In the wavelenths: 330-430nm,the unimodal shape of Skeletonema costatuma will turn into bimodal. In the wavelenths: 330-430nm,there are three peaks of the spectra of Synechococcus sp.And the intensity of the peak caused by chlorophylla at 430nm wavelenth will increase from day to day .But the influence only appear under the situation of 70001ux illumination intensity for Skeletonema costatuma and 20℃ for Synechococcus sp.2.Feature extraction methods were studied.Firstly the matrix made up by primitive spectrum data was processed by singlar value decomposition (SVD). the ratio value of S1 and ∑S was used to compare the similarities of spectra.According to the similar degree of spectrum,25 representative spectra was extracted by Principal Component Analysis.The 25 representative spectra can signify the essential feature of theabsorption spectra of the nine species. Among them. 4 spectra of Alexandrium tamarense,1 of Isochrysis galbana,2 of Platymonas helgolanidica,4 of Prorocentrum dentatumA of Skeletonema costatuma, 2 of Chaetoceros curvisetus,1 of Chaetoceros debilis Cleve,3 of Gymnodinium sp,and 4 of Synechococcus sp.Secondly.The wavelenths that have great ability to differentiate species were tested by three methods and 7 incontinuos wavelenths were received altogether.They are 340.5nm-420.5nm, 423.5nm-431nm,440.5nm-525.5nm,760.5nm.763.5nm,769.5nm and 856.5nm. Finally.the 7 incontinuous wavelenths of the 25 representative spectra make up the standard spectra.3. Cluster analysis and differentiate analysis were processed using the standard spectrum.In cluster analysis,the 3 standard spectra of Skeletonema costatuma of which the shape changed obviously and the 3 standard spectra of Synechococcus sp that changed obviously have remarkable difference to the other 19 spectra,and both of them have difference also.No matter how many numbers of the standard spectra there are of one species of the other 19 spectra.the similar degree of itself is higher than that between different species. But the standard spectra of Alexandrium tamarense, Prorocentrum donghaiense and Gymnodinium sp belonging to Dinophytav have great simility . The unimodal peak spectra of Skeletonema costatuma and the standard spectra of Chaetoceros curvisetus and Chaetoceros debilis Cleve,all of which belong to Bacillariophyta also have great simility.Furthermore,the similily is higher than that of Dinophytav.In differentiate analysis, the recognition correct rate was 80.86%.The recognition correct rate was 86.11% neglecting the discriminant error of the two species belonging to didymus genus.The recognition correct fate was more than 90% on division level.4. Qualitative and quantitative analysis of 61 mixtures of phytoplankton were carried out using the Non-negative Least Square (NNLS) method. The threshold value of 0.25 was chosen to assure the most abundant phytoplankton classes present can be identified and their relative abundance estimated, but the less abundant classes cannot. The recognition rate was 76%. The minimum of the determining range was 0.3mg chlorophyll-a/L. The recognition rate can be more than 90% if the content of themixtures were under the minimum. The abundance identified is at the same gratitude of the real abundance.In a word, this paper takes the preliminary research to the vitro absorption spectrum of the phytoplankton. Using the obtained characteristic spectra and NNLS method, dinoflagellates and diatoms can be distinguished from each other in laboratory cultures.