Sailing Seas of Plastic: Visualising 268,940 tonnes of plastic floating in the world’s oceans


By Timo Franz, Unfold Data (previously called dumpark) and Laurent Lebreton, The Ocean Cleanup

An estimated 5.25 trillion plastic particles weighing more than 268,940 tonnes are floating in the world’s oceans, according to a study published in the journal PlosOne and co-authored by dumpark co-founder Laurent Lebreton (now researcher at The Ocean Cleanup).

With the goal of turning the science into something everyone could relate to and thus raising awareness about the extent of plastic pollution, we produced the interactive map Sailing Seas of Plastic. In addition to the main study findings, we also wanted to exhibit the measurement data used to calibrate Laurent’s  previously developed numerical model and quantify floating, or more precisely buoyant, plastic concentrations in our oceans.

Modelling the relative distribution of marine litter at sea

A numerical model was initially developed by Laurent and his fellow researchers to simulate large scale oceanic transport of buoyant marine debris in the world’s oceans. The model successfully reproduced the formation of the five subtropical accumulation zones - also known as gyres - and identified several other accumulation zones in marginal seas such as the Mediterranean Sea. Further, it showed the relative contributions of the world’s different regions.

Driven by a 25 year simulation of global oceanic currents, the model tracked the pathways of virtual particles that were released into the model domain according to three source scenarios corresponding to inputs from:

  • coastal population,
  • inland population through rivers, and
  • the global shipping industry.

While this qualitative method allowed the researchers to discuss the formation of global scale accumulation zones for marine litter, without the availability of ground truth data the model would not provide a quantification of buoyant ocean plastic concentrations. In other words, the modelled distribution was dimensionless and only showed the relative distribution of a global integrated mass, representing 100% of the total available trash currently at sea.

These early findings were published in the Marine Pollution Bulletin, supported by an interactive visualisation: The Seas of Plastic.

Image: Relative distribution of buoyant ocean plastic, coastal release locations and principal ocean currents, screengrab from The Seas of Plastic.

Image: Relative contributions of different source regions to accumulation zones, screengrab from The Seas of Plastic.

Measuring ocean plastic concentrations at sea

A team of researchers from six countries, led by Dr Marcus Eriksen of The 5 Gyres Institute, collected plastic samples in the world’s oceans by trawling sea surface nets for small plastic particles and using systematic visual sightings for larger debris.

Video: 5 Gyres FAQ #5: How do we study microplastic pollution? from Chris Jones on Vimeo.

Plastic pieces collected at sea were categorised by size, counted and weighed. Numerical (pieces/km2) and mass (g/km2) concentrations were then calculated for the four different size classes considered in this study, ranging from microplastics smaller than grains of sand to macroplastics larger than plastic bottles.

Image: The four different size classes of ocean plastics considered in the study.

Overall, the team measured buoyant ocean plastic concentrations on 24 expeditions at 1571 locations, including 680 net trawls and 891 visual survey transects.

Image: Field locations where numerical concentration (pieces/km2) was measured by size class. Source: PlosOne.

Calibrating the numerical model

With access to measurements at sea, Laurent validated and calibrated his initial model, predicting numerical and mass concentrations, also known as count and weight densities (respectively in pieces/km² and g/km²), for each of the four size classes reported by Eriksen’s team, and ultimately estimating the global stock of buoyant ocean plastic.

Using his three different source scenarios, he fitted the numerical model predictions against measured data with a linear system of equations of the form:

Y = S * β + ε

where Y is a series of n observations, S a matrix of dimensionless model outputs for k source scenarios and n observations, and β and ε respectively k-long vectors of fitted coefficients and resulting error terms.

Image: Model results for global concentration of buoyant ocean plastic (pieces/km2) for the four size classes. Source: PlosOne.

Building the interactive

With the study accepted for publication, we decided to develop an interactive visualisation to communicate the findings to a larger audience. In addition, we also wanted to exhibit the 24 expeditions and the measurement data collected.

Visualising concentration

While investigating visualisation methods that would allow the audience to easily relate to the numbers, we found that dot density maps, commonly used to map the distribution of population, would achieve this very effectively. Rather than showing filled areas of estimated mass concentrations (expressed in kg/km2), a dot density map shows tangible items with an attributed weight (expressed in kg). Although not explicitly mapped, concentration gradients and patterns can easily be perceived from the resulting cloud of scattered dots.

Furthermore, dot density maps appear particularly well-suited, as they visually match the shown phenomenon of buoyant ocean plastic pollution establishing the more suitable metaphor of “plastic smog” over the common misconception of “trash islands”. After all, buoyant ocean plastic is not homogeneously distributed over continuous surface areas - like crude oil would be, for instance - and neither forms solidified islands, but it is composed of a wide variety of particles of different sizes and shapes dispersed over millions of km2 of water.

To transform the gridded data of mass concentrations to distributed dots, we calculated the number of 20 kg dots that would be contained, in each model cell (an area of 0.2 by 0.2 degrees or roughly 400 km²) where we randomly distributed them using a circular Gaussian function.

Image: Example areas with different mass and perceived mass concentrations.​

We then loaded the 13.4 million dots into Mapbox Studio Classic from which we could export a tiled basemap for various zoom levels, allowing to explore the data at different scales. Interestingly, earlier attempts to visualise dots of 1 kg (268 million dots) and 10 kg (26.8 million dots) failed due to software limitations.

Image: More than 13 million dots were plotted globally using Mapbox Studio Classic, screengrab from Sailing Seas of Plastic.

Breaking down findings by ocean

We created a bubble matrix chart to show the extent of plastic pollution by ocean and by size class. We used d3.js where circles represent total count or optionally mass.

Image: Ocean estimates by mass (tonnes), screengrab from Sailing Seas of Plastic.

Image: Ocean estimates by count (billion pieces), screengrab from Sailing Seas of Plastic.

Displaying field measurements

To support the model predictions, we also showed the various expeditions and corresponding measurement data from all 1571 locations. When in “expedition” mode, all expedition stages are plotted on the map from which they can be selected and explored.

Here, we decided to use heatmaps to depict measured plastic concentrations for every survey location (horizontal axis) and for the four debris size classes (vertical axis). The heatmaps are available both for numerical and mass concentrations (also referred to weight and count densities). Finally, to complement the visualisation of measurements and allow for comparison and model validation, an additional heatmap underneath shows the corresponding model estimates.

Image: Showing measured and modelled mass concentration (weight density) for one expedition stage, screengrab from Sailing Seas of Plastic.


The visualisation proved to be an effective method for communicating the scientific findings to a larger audience. In addition to the interactive reaching almost 200k visitors, the dot density map has and is still being frequently used in online and print media, researchers presentations, communication material of NGOs and even exhibitions (a smaller version of the dot map showing the Mediterranean Sea was printed and exposed over a few months during the "Vie d'Ordures" exhibition at the MUCEM in Marseille last year).

In subsequent research, supported by an interactive by The Ocean Cleanup, Laurent has looked at rivers as sources of ocean plastic pollution to help develop and implement mitigation strategies that prevent plastics from entering our seas in the first place.

About the authors

Timo Franz has a Masters of Business Information Systems from the University of Hamburg. In 2012, after a brief period of developing numerical modelling software, he co-founded dumpark (now called Unfold Data) with Laurent and Edith Woischin to create data visualisations and infographics for communicating data and complex information.

Laurent Lebreton graduated from Ecole Centrale de Nantes and has 10 years of experience in numerical modelling and physical oceanography. He started his research on the science of marine debris in 2010. Since then, he has authored and co-authored a dozen peer-reviewed publications on the subject, participated in multiple UNEP led working groups and is now internationally recognised in the field of marine debris dispersal research. He has joined The Ocean Cleanup in 2016 to assist the ocean plastic research team in understanding and quantifying plastic pollution in the oceans.

Explore the Sailing Seas of Plastic interactives here and here. Read the research they are based on here and here.