Tracking ocean eddies with AIS
Almost all ships nowadays are equipped with an Automatic Identification System (AIS), primarily to improve marine traffic safety. The system broadcasts signals containing the ship's identity, location, speed, heading and more every 2-10 seconds. These signals, collected and stored, make it possible to track the voyages of individual ships and to create time series of their movements. With the help of a simple parametric model, variations in the ship’s speed time series can be related to the wind, wave, and current conditions that are encountered during the voyage. One application of this model, to make forecasts of a ship’s speed in order to improve the arrival time estimate, was discussed earlier (http://www.melodiesproject.eu/content/how-open-data-can-improve-ship-arrival-times). Here we consider the potential of using AIS time series to assess surface currents, turning ships into moving buoys.
Currents worldwide are forecasted by ocean circulation models. Below, the results from two state-of-the-art models for the notoriously complex Agulhas currents are displayed.
On a large scale, the two models show similar patterns but in detail large differences are found. An important reason for the limited accuracy of ocean circulation models is the scarcity of current measurements. Ship’s AIS data could be a valuable addition to the available buoy and drifter data, and the satellite altimeter measurements of the sea level.
An analysis of a ship’s speed variations gives an assessment of the surface current along the direction of the ship’s course, and the deflection from a straight course are used to assess the cross currents. The AIS time series are filtered to remove noise and outliers. Corrections for wind and wave effects, using model data, are generally minor. The analysis assumes the absence of manual speed or course corrections by the captain, but some manoeuvres can be detected in the AIS data and corrected for. As only current variations can be deduced with this method, it is best applied in areas with strong variations, either in time (e.g. strong tides) or in space (e.g. eddies). The latter is the subject of this blog.
In a case study an eddy was selected off the coast of S.E. Africa. The figure below shows the currents pattern, obtained from the MyOcean model. The current speed is given in m/s by the colour and the arrows show the direction and magnitude. AIS data were collected from four ships that sailed through this eddy within a time frame of 5 hours; the dots show their position.
Ship 3 (the black dots) sailed almost straight through the centre of the eddy. The figures below display the along ship and cross currents obtained from the model (black dots) and from the AIS data (red dots). The direction of along ship currents is 245 degrees (going to) and of cross currents is 155 degrees. To calculate these currents two course corrections and one speed correction were deduced from the AIS time series. The area with strong cross currents is narrower in the AIS data but for the rest there is a remarkable agreement between the model and assessed AIS currents.
Ship 2 (blue dots) sailed a more northerly track. In the figure below the along ship currents are displayed, showing a difference between the model results and the AIS assessments. The latter indicate a narrow area with strong currents whereas the model gives a much smoother pattern. An analysis of the course variations suggest many small course corrections, which makes an assessment of the cross currents less reliable.
The ships 1 (green dots) and 4 (red dots) sailed still further to the north, closer to the coast. Ship 1 sailed towards the north and was slowed down by the currents while ship 4 sailed towards the south and was accelerated. The model and AIS along ship currents are displayed below. For easier comparison the along direction is 245 degrees for both ships. In these plots a clear difference between model and AIS results is seen: the AIS data from both ships show an area with strong currents between 31 and 31.5 degrees longitude that is absent in the model. It is unlikely that these similar patterns are caused by ship’s manoeuvres, especially since the ships sailed in opposite directions.
The three ships that sailed north of the eddy’s centre consistently and independently show an area with strong currents. Where the model finds a gap in the northerly part of the eddy’s ring, the AIS data suggest that the strong currents there continue. Of course, this case study is too limited to draw strong conclusions from. But it confirms the potential of AIS data as a new and valuable source of surface current assessments that can help to improve the accuracy of the ocean circulation models. And moreover, AIS data is freely available, waiting to be analysed.
More details on this case study can be found in the graduation thesis of Ruben van der Neut.