Data on fishing activity out at sea has traditionally been imprecise, difficult to access, and spread between many different regulating authorities. With the publication today of “Tracking the Global Footprint of Fisheries” in Science and the release of a public dataset of global fishing effort we hope to enable researchers and fisheries managers to fully take advantage of AIS tracking data for ocean conservation. (Read this excellent article on the work in The Washington Post.)
The data and analysis presented in this paper have been the result of a long-term collaboration between researchers at SkyTruth, Global Fishing Watch, Google, and universities in the United States and Canada. The research has been led by David Kroodsma, research program manager at Global Fishing Watch. Other authors from the GFW and SkyTruth teams are Paul Woods, CTO of GFW, Nate Miller, SkyTruth research analyst, Tim Hochberg, machine learning engineer at GFW, and myself, an analyst at SkyTruth. Along with other academic researchers we have worked to characterize the population of vessels broadcasting AIS and to assess the limitations in AIS coverage and reception.
Central to the work being presented is a description of the data pipeline and modeling used to process the vast quantity of AIS data broadcast by the over 70,000 vessels now tracked in Global Fishing Watch. Machine learning was used to classify the tracks of these vessels and infer both where they were fishing and what type of fishing gear they were likely using. Based on vessel movements, models could even predict vessel characteristics like length and engine power.
The temporal and spatial precision of this new global fishing effort dataset highlighted some surprising regional variations. Weekends are often taken off by fishermen in Europe and North America. This is not the case on the Chinese coast where fishing is only interrupted by the Chinese New Year and a summer fishing moratorium. This can seen by comparing Chinese and non-Chinese fishing vessels in this data visualization from Global Fishing Watch.
The most distinct spatial patterns of fishing effort can be seen to result from differences in regulation. More subtle effects are seen from variables like sea surface temperature and net primary production. Below, you can see 2016 fishing effort off the coast of Patagonia, which shows both intense fishing activity by foreign vessels just outside the EEZ boundary and a checkerboard pattern within the EEZ due to Argentine regulation of the hake fishery.
The data appearing in the image above is part of the public dataset that is being released along with our paper. Researchers can select maps for different regions or fishing gear types and also download the raw data underlying the images. To learn more about the study and to access the data, click here.
This publication and data release is a milestone for our analysis of the global AIS dataset for fishing vessels but we still have a lot learn about the patterns of vessel movements we have characterized here. We hope our work can spur an increase in the use of AIS tracking data for fisheries research and regulation and we look forward to working with more partners to better understand this new data resource for marine conservation.