The algal bloom monitoring buoys being installed in Gulf St Vincent

The first of two algal bloom monitoring buoys were installed in Gulf St Vincent at St Kilda and Stansbury during December. A total of 6 of the buoys will be installed. According to https://www.algalbloom.sa.gov.au/news/algal-monitoring-buoys-ahoy , “Six hi-tech monitoring buoys which will automatically detect future harmful algal blooms before they reach our shores are being deployed across South Australian waters. The state-of-the-art oceanographic moorings are equipped with specific sensors for early detection of key harmful algal bloom (HAB) species such as Karenia. If a HAB species is detected, an immediate alert is sent to scientists via satellite communication. They also provide real-time, continuous monitoring of critical parameters including temperature, salinity, chlorophyll, dissolved oxygen and turbidity. Part of the State and Federal Government’s $102 million Summer Plan, the sophisticated monitoring system strengthens South Australia’s ability to detect, track, research and forecast coastal water quality and harmful algal blooms. The moorings will form a core component of the South Australian Integrated Coastal Water Quality Monitoring and Forecasting System, a comprehensive scientific program led by the new Office for Algal Bloom Research in conjunction with CSIRO. The monitoring system builds on the existing Integrated Marine Observing System moorings which have been in place in open waters off the coast since 2008, with the new moorings enabling coverage in inner coastal areas. The first two monitoring buoys are being installed at St Kilda and Stansbury in Gulf St Vincent today (18 Dec 2025), with the remaining four sites to be finalised in consultation with industry.”

Delta Environmental Consulting reported on their Facebook page about having been asked lots of questions about algal monitoring buoys that use automatic image recognition. They say that the technology is in its early stages yet, adding, “Imaging FlowCytobots are a big step up from automatic particle counters, or even from particle counters that detect the various chlorophyll reflectances, both of which have been round for a while. The former simply counts microscopic particles. The latter sorts them into green algae, Cyanobacteria, chromophytes and non-phytoplankton debris. But the dream of automatic identification to species level has proven to be only reachable to a certain extent. It IS happening, but not “off the shelf”. You buy the hardware then employ a team of algal taxonomists and digital image database programming specialists to build a dataset of all the local plankton species.  Images from all angles and in all states of disarray. Not just the image of you from the school photo – the one of you coming home from the Melbourne Cup as well. These datasets contain hundreds of thousands of carefully identified images, maybe millions in very biodiverse localities. An example of a current project to build one of these datasets is the Mediterranean plankton dataset, linked here (https://szn-ifcb.weebly.com/what-is-the-ifcb.html?fbclid=IwY2xjawOzB0NleHRuA2FlbQIxMQBzcnRjBmFwcF9pZBAyMjIwMzkxNzg4MjAwODkyAAEemdkhO4ZDhdy5zyWRZ88kiac8oxnwRtwm7l8jzpV0bdmDRM-cJXvwFA8Pw6Q_aem_DGoAtkqX1zps0agZpF7lzw ).

The reason I suspect that this technology will be incorporated, eventually, into the buoys being deployed in SA Gulf waters? Firstly, because we have bought the very expensive buoys – when satellite sensors already tell us the basics of salinity, water temperature and chlorophyll reflectance, if not as perfectly precise as we would like. And secondly, on SEEK there are quite a tranche of jobs on offer in exactly the right categories (taxonomists, image processing, IT database programming)  at the new Office for Harmful Algal Blooms based at SARDI.”

Faith Coleman commented, “the buoys will be using hyper spectral sensors to apply indexes for known species/species groups, as well as chlorophyll, but the power consumption is so high that it is expected that the AI species identification system described will only (most likely ever) be usable for the fixed location systems – i.e. those attached to jetties or intakes. They also require a huge amount of upkeep, so need to be accessible. At present I believe there (are) plans for one AI system, but not usable in earnest for some years yet.”

There are many more comments following Faith’s.