Application Of Expectation Maximum Clustering Analysis Model Of Biogeochemical Characteristics In The Mediterranean Sea

The Mediterranean Sea is the typical epitome of the global ocean,because the biogeochemical characteristics diverse highly over the whole basin and it covers all three main ecological regions:Bloom,Intermediate Bloom,and Oligotrophic Desert.There is no doubt that studying the Mediterranean biogeochemical characteristics is important for us to understand the whole ecological geography of the global oceans.This dissertation mainly focuses on three aspects describing as follows:Firstly,using satellite remote sensing data and model data to analyze the temporal and spatial distribution of the overall biogeochemical characteristics of the Mediterranean Sea,including topography,circulation distributions,sea surface temperature(SST)a sea surface salinity(SSS)and sea surface density distributions and the distribution of the mixed layer depth,as well as nutrient variations(N,P,N:P).Secondly,by using the polynomial mixed EM clustering analysis model,we have drawn three representative ecological partition of the Mediterranean Sea:the bloom region(A1),the intermediate bloom region(A2),oligotrophic desert(A3).Then,we have analyzed seasonal cycles of the 3 phytoplankton physiological parameters(Chla,phytoplankton carbon and the ratio of both:Chl:C),The effects of photoacclimation on these physiological parameters were also described.Because of the photoacclimation,the seasonal cycle of Chla and carbon are not always in coincidence,the seasonal pattern of Chla and Chl:C is spring>winter>autumn>summer,the seasonal pattern of carbon is spring>winter>summer>autumn.Finally,we have described the interannual variations of phytoplankton physiological parameters(Chla,phytoplankton carbon and their ratio Chl:C)in the 3 partitions mentioned above(Al,A2,A3)by using wavelet analysis.We have also analyzed their relationships with environmental factors and climatic factors,the El Nino Southern Oscillation Index(ENSO)and the North Atlantic Oscillation.Overall,the correlation of phytoplankton parameters with ENSO,SST,light intensity is remarkably negative,with MID is remarkably positive.As for the NAO index,there is a little bit difference between A3 and Al,A2,the correlations of NAO index with phytoplankton parameters in A2 and A1 are negative but in A3 is positive.There are 3 innovative points of this dissertation as follows:The first is the application of Expectation Maximum clustering method which is better than K-means for the clustering smoothing and denoising.The second is using combined phytoplankton carbon biomass and Chla as the clustering input data instead of the Chla only which can avoid the misunderstanding between the Chla cycles and the actual phytoplankton biomass variations.The third is when doing the interannual analysis of phytoplankton phenology,we combined their relationships not only with climatic index but also the environmental factors.