Study On Algorithms For Phytoplankton Blooms Identification Using Satellite Remote Sensing

Remote sensing of marine phytoplankton is of great ecological significance because ocean plays a big role in global carbon cycle.As a start of this thesis,we tried to use some empirical formulas to modify MODIS Rrs(412)and Rrs(443)which can be problematic in case two waters.Then corrected Rrs were applied to revised algorithm for dinoflagellate and diatom blooms identification in East China Sea.We also did some further research about the universal application capability in global ocean and indentification possibility for various blooms.Some detailed results are as follows:We found the correlation formula between Rrs(412)Rrs(443)and Rrs(488)by statistical analysis of NOMAD data and field data in China seas collected by scientific cruises.After evaluation of MODIS Rrs through certain score schme,those pixels lower than 0.8 were corrected by those formulas.The results of Taiwan strait and gulf of Mexico indicate that this correction method is efficient with not only valid Rrs increased after correction,but also the invalid retrivels of QAA and GSM model decreased significantly using corrected Rrs.In China seas,diatom and dinoflagellate blooms account for the vast majority of bloom cases.Both Shang and Tao proposed new algorithms for the two kinds of blooms identification.With the help of in situ data of some bloom cases we made some revisement to the two algorithms.Revised algorithm was applied to MODIS L2 data to monitor the dinoflagellate and diatom blooms in East China Sea from 2003 to 2015.The results suggest that blooms were serious for every year since 2003 with features like high frequency and large covering area.Dinoflagellate blooms are more frequent and serious than diatom blooms in ECS.The diatom blooms tend to decreasing for a long time and dinoflagellate blooms show siginificant decrease after 2013.This algorithm is also applicable in some other seas like Bohai Sea,offshore of South Korea,North America,South Africa,but failed in offshore of Chile because this algorithm needs satellite data of high accrucy as inputs.With the analysis of field data or corresponding satellite data we believe that blooms dominated by Phaeocystis globose can be identified using certain method or criteria on condition that more field data are provided for algorithm accuracy improvment.As for the blooms caused by Aureococcus anophagefferens,they can be detected by our algorithm but difficult to differentiate them from dinoflagellate or diatom bloom waters with remote sensing method.