Forecasting sargassum across the tropical Atlantic with SARTRAC-EFS

The emergence and spread of sargassum seaweed in the tropical Atlantic is a major environmental challenge for many communities of the islands and coastal states of this region. Observed in satellite images, several groups have worked to predict the movement of sargassum, subject to ocean currents and surface winds. We add a little bit of biology too!

Through the SARTRAC project , we developed an ‘Ensemble Forecast System’ (EFS), hence SARTRAC-EFS. We use SARTRAC-EFS to forecast the drift of sargassum on timescales of 3-6 months, and its arrival offshore at selected locations. Sargassum is represented with many thousand ‘particles’, in proportion to how much sargassum is ‘seen’ by the satellite. Individual particles are then moved with sampled ocean currents, and optionally winds, for up to 6 months. Each particle follows a unique ‘trajectory’. We average all the trajectories together to obtain predicted distributions and local forecasts.

A full description of SARTRAC-EFS can be found in:

Marsh R, Addo KA, Jayson-Quashigah P-N, Oxenford HA, Maxam A, Anderson R, Skliris N, Dash J, Tompkins EL. Seasonal Predictions of Holopelagic Sargassum Across the Tropical Atlantic Accounting for Uncertainty in Drivers and Processes: The SARTRAC Ensemble Forecast System. Front. Mar. Sci. 2021; 8:722524. doi: 10.3389/fmars.2021.722524

Reflecting on the challenges of this forecasting, we also evaluate such approaches in:

Marsh R, Oxenford HA, Cox S-AL, Johnson DR, Bellamy J. Forecasting seasonal sargassum events across the tropical Atlantic: Overview and challenges. Front. Mar. Sci. 2022; 9:914501. doi: 10.3389/fmars.2022.914501

Here, we provide some examples from SARTRAC-EFS, with predicted distributions in the Caribbean and West African regions of the tropical Atlantic, along with corresponding forecasts around Jamaica and along the coast of Ghana.

First choose to look at 30-day snapshots of sargassum distributions (percentage cover) in the Caribbean or West African region. For the Caribbean region, you will see distributions up to 180 days from the initial satellite-observed distribution. For the west African region, these distributions are only reliable up to 90 days. In both cases, we show distributions subject to ocean currents alone, and subject to ocean currents plus the additional effects of wind acting on exposed sargassum (sometimes called ‘windage’).

Carribbean

West Africa

The four forecasts are obtained for different assumptions about winds and biology:

Upper left: drift with current only
Upper right: drift with current plus winds
Lower left: drift with current only, subject to growth and loss
Lower right: drift with current plus winds, subject to growth and loss

We estimate growth and loss (mortality) along each particle trajectory, according to surface temperature. Sargassum grows fastest at 26-27oC, but increasingly dies off at temperatures above 29oC. Through recent laboratory and mesocosm experiments, we now know more about the dependence of growth on temperature, in particular that the ‘new’ sargassum grows well at higher temperatures than previously thought, so these results may need updating.

Notice how variable the forecasts are, depending on the assumptions about winds and growth/loss. Notice also that each forecast is based on a number of ‘ensemble members’ that we average together (the thick purple line), accounting for considerable uncertainty in wind, currents and temperatures, for any particular year – a bit like the weather forecast.