In recent years, Japanese estuaries and brackish lagoons tend to experience phytoplankton blooms of dinoflagellate in winter. In some lakes and estuaries, yearly maximum chlorophyll a (Chl.a) concentration was observed in winter season, however, the mechanisms of such blooms are poorly understood. In this study, dynamics of Chl.a of the Kiso and Nagara River estuaries was investigated, by using long-term monitoring data set of water quality during 1995-1998. In general, the winter phytoplankton bloom outbreaks in February until the middle of March, when river flow rates abruptly increase by the snownelt runoff. The weather during the bloom season is relatively calm and the flow rates of the rivers are low and stable. The chloronity (Cl concentration) was found to be a primary component to govern the temporal variation of the Chl.a concentration in the surface layer of the two estuaries. As Cl varied through the tidal motion, Chl.a concentradon fluctuated not only with the semidiurnal/diurnal tidal cycle but also with the spring-neap-spring tidal duration. Maximum Chl.a was observed at around 10,000 mg/l in Cl.
　A simple phytoplankton growth model was constructed to invesdgate the mechanism of the winter bloom. The model includes phytoplankton growth as a function of Cl, light intensity, nutrient (DIN), and water temperature. Respiration and mortality terms are also incorporated as a function of temperature. The model was firstly applied to the Kiso River estuary. The model calibrated by using the 1996 data set collected at an observational station in the Kiso River estuary, then validated against 1997-1998 data sets of the same station. The model well reproduced not only the temporal variations from diurnal to spring-neap-spring cycles in 1996, but also year-to-year variations in 1997 and 1998. Relatively large and continuous outbreak of the bloom was observed in 1997, whereas the bloom was intermittent and the Chl.a concentration was low in 1998. Tne model analysis indicates that the difference of the bloom in the two years was caused by difference in Cl concentration. The low chl.a in 1998 was due to low Cl caused by abnormally high discharge of the river. The model was also applied to the data sets in the Nagara River estuary in 1997 and 1998. The model calculations satisfactorily reproduced not only year-to-year variation (relatively high Chl.a in 1997 and low in 1998) but also spatial difference between the two observational sites located in the Nagara River estuary, in case that nutrient limitation was incorporated. However, the model poorly reproduced the spring-neap-spring variation in Chl.a, as the Nagara River estuary is physically more complex due to the presence of the rivermouth barrage and inflow of the Ibi River. The model analysis also indicates that grazing pressure by zooplankton is higher in the Nagara River than that in the Kiso River, which may be due to stagnant character of the water body, associated with the presence of the Nagara rivermouth barrage.