Summer 2015, as the eighth month of the year approaches (August), we are starting to hear and read in the news about sudden algal blooms occurring in freshwater lakes and marine environments. Many popular beach resorts have put up warning signs proclaiming the water is unsafe for recreational purposes.
This is an understandable precautionary measure since phytoplankton abundance or more specifically the cyanobacteria blooms also known as harmful algae blooms (HABs) may cause serious health threats to humans, animals, and may disturb the overall water ecosystem.
Some of the cases reported this summer include the harmful algae bloom problem in Lake Erie, the smallest by volume of the Great Lakes. Another report from the Iowa State, referring to Crystal Lake, says that a record number of algal toxin warnings (25) were published by the middle of August this year and yet the summer season has three more weeks to go. There was also the unexpected event of the red algal tide that occurred on the coast of the Netherlands between Katwijk and Scheveningen in only a day.
There are plenty of harmful algae bloom related cases but what is evident is that their number elevates with each year. The immediate response when a bloom happens is common for all of the above-listed cases and follows the same pattern. The procedure includes water sampling and algae analyses. Normally, by the time that a government institution comes out with an official warning report, the bloom has already started to decrease while the warnings remain for a certain amount of time as a precautionary measure.
Accordingly, a question arises: “Can we predict these harmful algal blooms and how can we be one step ahead in order to react before the ‘damage’ is done?”
The most rational answer is to create better policy control of nutrient loads (such as N and P) that end up in freshwater or marine water bodies causing highly eutrophic conditions and algal blooms consequently. However, this is a long-term solution dependent on regulatory measures that normally take years before enforcement and even longer for any visible improvements. Hence, it is more important to check the options on what can be done in the short term as an early warning measure.
Real-time monitoring of water quality parameters related to phytoplankton dynamics such as Chlorophyll-a, Temperature, DO, pH, Turbidity is an essential tool for short-term forecasting of potentially harmful algae blooms. Moreover, online water quality monitoring is an important part of the European Water Framework Directive (WFD).
Dealing with Water Quality Analysis on a daily basis and looking at the amounts of real-time monitoring data, I decided to conduct a small research analysis to see how I could use the in situ parameter readings to hypothetically predict these algal blooms up to one week ahead of their occurrence.
There are several developed models with most of them based on algorithms that commonly make use of in situ real-time monitored water quality data derived from deployed sensors probes. For instance, in the case study of Lake Enghien (France), repeatedly threatened by blooming of Planktothrix agardhii, a model for short-term prediction of these cyanobacterial species and the overall cyanobacteria dynamics has been used. The model is a neural network of Non-linear Auto Regressive and eXogenous inputs (NARX) that use the Chlorophyll-a concentrations, the water temperature of the previous 3 days and atmospheric temperature of the next 3 days. By implementing these parameters, the model can forecast the growth of Planktothrix agardhii up to 3 days ahead (Silva et al., 2011).
Another more recent project describes the case of Three Gorges Reservoir in Xiangxi Bay (China), where a 1-4 days prediction model of phytoplankton bloom has been successfully developed by using continuous monitoring of Chlorophyll-a concentration, which are later integrated into hybrid evolutionary algorithms (HC and DE). However, this model managed to precisely predict an early warning hypertrophic conditions (Chl-a 8-25 µg/L) only 1-2 days ahead while for good to eutrophic conditions the model could precisely predict 1-7 days ahead (Ye et al., 2014). It is important to mention that in both cases, other monitored physical and chemical parameters such as pH, DO, Turbidity, conductivity, water temperature as well as meteorological elements (air temperature, wind speed and rainfall) were considered or integrated in these prediction models.
Similarly, at LG Sonic we have developed an online water quality monitoring software (MPC-View) where we have real-time overview of the essential parameters for algae prediction, such as Chlorophyll-a, Phycocyanin, DO, pH, Temperature, and redox. The data derives from deployed MPC-buoy system equipped with monitoring probes and ultrasonic algae treatment devices. After we have identified the specific algae genus present in the water, we can set the correct ultrasonic program and target the algal bloom before it occurs.