Real-time modelling and prediction of harmful algal blooms

‘Real-Time Modelling and Prediction of Harmful Algal Blooms’ is one of a suite of SAIC-supported projects in the crucial area of harmful algal blooms and their impact on Scottish fish farming. This also includes:

• a costed extension of the original project: ‘Real-time modelling and prediction of harmful algal blooms to minimise their impact on finfish aquaculture: high-frequency automated monitoring’; 
• and a project funded under SAIC’s R&D Boost call: ‘High-resolution modelling of the Shetland Isles to safeguard finfish gill health’.

In recent years, the Scottish salmon industry has been negatively impacted by an apparent increase in the number of coastal harmful algal bloom events (HABs). These HAB events have resulted in challenges to gill health of the farmed salmon, causing increased mortality and overall decline in production efficiency. The negative effects on the gills of fish caused by HABs can include physical damage and/or exposure to toxins produced by the algae. This can result in an acute mortality event, or the development of chronic gill pathology leading to decreased respiratory efficiency, or in combination with other waterborne pathogens can develop into multifactorial complex gill disorder (CGD).  

The seasonality of HABs often coincides with periods of decreased dissolved oxygen due to increases in seawater temperature, which can have a compounding effect for impaired gill health. The deterioration in gill function can also lead to increased risk of mortality if the farmed fish population are subsequently required to undergo an intervention to treat other challenges to fish health, such as for amoebic gill disease (AGD) or for the control of sea lice. Gill health has been prioritised as one of the main risks to farmed fish health in the Scottish Government Framed Fish Health Framework (FFHF) and a key area where improvements are expected to further enhance fish health and welfare, and to increase farmed fish survival.  

Mitigation of the effects of HABs on farmed fish will therefore support the sustainable development of the salmon sector. However, as HABs are primarily natural events that typically develop offshore before moving towards the coastline, their prevention is not a control option. Therefore, improved mitigation of the impacts of HABs requires advance warning of the timing and location of blooms before they reach a farm. This will allow husbandry strategies to be implemented on site to protect fish health.

As HABs are transported by relatively well-understood oceanographic currents, it is possible to use mathematical models to predict the development and trajectory of the blooms. This project will therefore develop a web-based alert system that will, for the first time, use phytoplankton cell counts undertaken at fish farms alongside satellite remote sensing products to initiate high-resolution physical-biological mathematical models of the Scottish west coast. Running these models will generate predictions of the future (~ next five days) location and timing of HAB events. These model-generated early warnings will be immediately available to stakeholders via the existing web portal www.HABreports.org (in a similar manner to a weather forecast), allowing mitigation action to be taken before harmful algal bloom impacts occur. 

This is a costed extension of our project to develop a methodology to record, report and predict (using mathematical modelling) the appearance of fish-killing HABs in Scottish waters to provide maximum early warning of harmful events and allow the aquaculture industry to implement husbandry strategies to protect fish health.

In summary, the project has two main aims:

1. To develop a data reporting, recording, visualisation and alert system (HABreports.org) for harmful phytoplankton cell counts made at salmon aquaculture sites in Scotland (specifically Scottish Sea Farms and Mowi within the project, but with a view to extending this more widely post-project).
2. To use the phytoplankton cell counts made at fish farms to initiate a high-resolution physical-biological mathematical models of the Scottish west coast. Running these models allows us to generate predictions of the future (next five days) location and timing of potentially fish-killing HAB events.

At the end of the extended project, we expect to have raised the TRL level of the alert system from 7 to 9 allowing its routine operation to support the Scottish aquaculture industry.

The Scottish salmon sector is being negatively impacted by an apparent increase in the number of coastal harmful algal blooms (HABs). A harmful cell’s first port of contact is the gills, where it can cause physical damage or exposure to algal toxins. Gill function is also reduced due to dissolved oxygen depletion by HABs. Effects on the fish include acute mortality and chronic gill pathology. This can develop into complex gill disease (CGD) and decreased respiratory efficiency.

Gill health has been prioritised as a major risk to farmed fish health in the Scottish Government Framed Fish Health Framework and is an area where innovation is required to improve fish health and welfare and increase farmed fish survival. The financial implications of HABs can be significant, with bloom events costing hundreds of thousands of pounds.

Mitigation of the effects of HABs requires warning. Currently, this is achieved by daily collection of water samples at farms, with microscope analysis to enumerate harmful taxa. However, such monitoring acts only as a “nowcast” rather than a forecast, and does not provide early warning of developing gill disease events or sufficient time to take action.

As most HABs occur naturally and are transported by well-understood oceanographic currents it is possible to use mathematical models to predict their development and trajectory. Model results can then be reported to aquaculture practitioners through a web-based alert system.

As part of an ongoing SAIC-funded project – ‘Real-time modelling and prediction of harmful algal blooms to minimise their impact on finfish aquaculture” (SAIC-HABs)’ – we have developed such a model alert system for Scottish waters. Within this, we trialled two modelling systems. The high-resolution WeStCOMS unstructured grid model system developed at SAMS (that can resolve specific locations to 10s of meters), and the 1.5km grid Met Office AMM15 model. We found that the lower resolution of the latter model resulted in it being unable to adequately simulate bloom dynamics.

The primary objective of this “boost” project is therefore extending the domain of the high-resolution WeStCOMS model from the Scottish west coast and islands to include the Shetland islands, and hence to provide HAB alerts for all of Scotland’s marine aquaculture.

The project will also allow us to evaluate the potential to initiate the model alerts with data from the two Imaging FlowCytoBots that we have recently deployed, with SAIC funding, in Shetland. These instruments use imaging and AI technology to enumerate phytoplankton cells. The water column can be surveyed multiple times per hour with results being reported in real-time on a web interface.

Finally, it is important to recognise that poor gill health of farmed fish has a range of causes. Early prediction of gill disease outbreaks would enable early actioning of mitigation measures. The accuracy of such predictions will be improved by including real- and modelled phytoplankton data. Therefore, combining oceanographic phytoplankton dispersion models with predictive gill disease models will set the foundations for a future decision support tool that can provide early warning for marine gill disease Scotland-wide.