Phytoplankton sampled weekly within the IGA monitoring programme since 1989, corre- spond to one of the longest time series in France. This time series is significant to further knowledge on phytoplankton bloom dynamics in the context of the regional massive outburst of the Prymnesiophycea Phaeocystis globosa. We provide a four steps innovative method of missing data completion and classification applied to Chlorophyll-a (Chl-a) time series in order to define various typology of algae blooms. These scheme are then analysed according to taxonomic composition. First step corresponds to data preprocessing: data validation, time shifting and missing data completion using fuzzy logic. Second step consists of identifying characteristic frequencies for multi-scale modelling. In our Chl-a case study, conventional approaches like correlation techniques, trend or mode decompo- sition allow to extract main cycles when considering phytoplankton blooms. Nevertheless, they are ineffective for analysis at small scales, closer to the real temporal dynamics of phytoplankton blooms. Spectral classification is used to identify sets of yearly shape-profiles. Five Chl-a pro- files were discriminated, in which 2 are dominant, non successive and with a non-regular rotation. Third part concerns the identification of recurrent or rare events that look like an asymmet- ric bell-shaped curve. As first study, we consider these events as Gauss curve. So, an algorithm to estimate gaussian mixture is proposed to model seasonal series and each yearly Chl-a series. A comparison between model from the seasonal decomposition and obtained models per year show an important variability of the number of gaussians, of their shapes and also of their dates of beginning and end of the bloom, duration. Last process is the correspondence analysis between obtained Gauss curves and the associated phytoplankton composition. Three gaussians are characterized by one to three taxa (> 95% of abundance) and a dominance of Phaeocystis Globosa. Other gaussians are relative to Bacillar- iophyceae dynamics.