Real time observations of coastal algal blooms by an early warning system

In eutrophic sub-tropical coastal waters around Hong Kong, phytoplankton or unicellular microalgae can grow rapidly to very high concentrations under favourable environmental conditions. These harmful algal blooms (HABs) have led to massive fish kills, hypoxia, and beach closures. However, to date the causality and mechanism of coastal algal blooms are still poorly understood. A remotely controlled autonomous real time field monitoring system has been developed to continuously track the changes in chlorophyll fluorescence, dissolved oxygen and other hydro-meteorological variables at two representative mariculture zones. The system can give an alarm when a bloom is detected, so that timely manual water quality sampling can be carried out to supplement the telemetric data. During 2000-2003, the system has successfully tracked 19 algal blooms. In the shallow weakly flushed coastal water (depth 7-10 m, tidal current 5-19 cm s(-1)), the bloom is short-lived, typically lasting a few days to over a week, with chlorophyll and DO concentrations in the range of 20-40 mg m(-3) and 2-15 g m(-3), respectively. It is found that: (1) the chlorophyll concentration is strongly correlated with its past values in the previous week, suggesting an auto-regressive type of algal dynamics; (2) the dissolved oxygen can reach highly super-saturated levels (12 g m(-3)) during a diatom bloom, and decreases to below 4 g m(-3) at the tail of the growth phase; (3) in contrast, a dinoflagellate bloom is characterized by a much more pronounced vertical structure. Diel vertical migration and aggregation to dense layers are clearly observed. Significant dissolved oxygen consumption is associated with the migration, resulting in DO drops by as much as 6 g m(-3) during the bloom; (4) the predominance of diatoms and dinoflagellates at the two sites can be explained in terms of the different hydrographic and nutrient conditions (the N:P ratio). Net algal growth rate, sinking and swimming velocities are derived from the in situ bloom data. The 4-year high frequency data set provides a basis for development of models for forecast of harmful algal blooms. (c) 2005 Elsevier Ltd. All rights reserved.