Improving Fishing Pattern Detection from Satellite AIS Using Data Mining and Machine Learning

RESEARCH ARTICLE Improving Fishing Pattern Detection from Satellite AIS Using Data Mining and Machine Learning Erico N. de Souza1☯*, Kristina Boerder2☯*, Stan Matwin1,3, Boris Worm2 1 Big Data Analytics Institute, Faculty of Computer Science, Dalhousie University, Halifax, NS, Canada, 2 Biology Department, Dalhousie University, Halifax, NS, Canada, 3 Institute of Computer Science, Polish Academy of Sciences, Warsaw, Poland a11111 ☯ These authors contributed equally to this work. * (ENS); (KB) Abstract OPEN ACCESS Citation: de Souza EN, Boerder K, Matwin S, Worm B (2016) Improving Fishing Pattern Detection from Satellite AIS Using Data Mining and Machine Learning. PLoS ONE 11(7): e0158248. doi:10.1371/ journal.pone.0158248 Editor: Athanassios C. Tsikliras, Aristotle University of Thessaloniki, GREECE Received: February 11, 2016 Accepted: June 12, 2016 Published: July 1, 2016 Copyright: © 2016 de Souza et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: Raw S-AIS data can be obtained commercially from provider exactEarth through their website (http://www.exactearth.com/ products/exactais). A key challenge in contemporary ecology and conservation is the accurate tracking of the spatial distribution of various human impacts, such as fishing. While coastal fisheries in national waters are closely monitored in some countries, existing maps of fishing effort elsewhere are fraught with uncertainty, especially in remote areas and the High Seas. Better understanding of the behavior of the global fishing fleets is required in order to prioritize and enforce fisheries management and conservation measures worldwide. Satellite-based Automatic Information Systems (S-AIS) are now commonly installed on most ocean-going vessels and have been proposed as a novel tool to explore the movements of fishing fleets in near real time. Here we present approaches to identify fishing activity from S-AIS data for three dominant fishing gear types: trawl, longline and purse seine. Using a large dataset containing worldwide fishing vessel tracks from 2011–2015, we developed three methods to detect and map fishing activities: for trawlers we produced a Hidden Markov Model (HMM) using vessel speed as observation variable. For longliners we have designed a Data Mining (DM) approach using an algorithm inspired from studies on animal movement. For purse seiners a multi-layered filtering strategy based on vessel speed and operation time was implemented. Validation against expert-labeled datasets showed average detection accuracies of 83% for trawler and longliner, and 97% for purse seiner. Our study represents the first comprehensive approach to detect and identify potential fishing behavior for three major gear types operating on a global scale. We hope that this work will enable new efforts to assess the spatial and temporal distribution of global fishing effort and make global fisheries activities transparent to ocean scientists, managers and the public. Funding: This work was partly funded by the NSERC CREATE Transatlantic Ocean System Science and Technology (TOSST) and the National Science and Engineering Council of Canada. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. 1 Introduction Competing Interests: The authors have declared that no competing interests exist. A common challenge in ecology is the mapping of dynamic patterns of human activity across vast areas in order to understand and track their ecosystem impacts on regional and global scales PLOS ONE | DOI:10.1371/journal.pone.0158248 July 1, 2016 1 / 20 Identifying Fishing Activity from AIS [1–3]. While important from a scientific perspective, there are also many other obvious applications, including the monitoring of marine fisheries and the enforcement of spatial management measures, such as marine protected areas (MPAs), ecologically and biologically sensitive areas (EBSAs) as well as fisheries closure zones. While the reception range of coastal monitoring tools such as tower-based applications (tAIS, radar) is limited to inshore areas, long-range tools such as satellite-based AIS (Automatic Identification System) and VMS (Vessel Monitoring System) provide insight into vessel movements elsewhere. VMS was specifically designed to monitor commercial fisheries while AIS was intended as a safety feature to avoid vessel collisions under low visibility. While the use of VMS devices is mandated only for some fleets in individual nations, the International Maritime Organization (IMO) has made the carrying of an AIS transponder mandatory for all vessels larger than 300 gross tons or carrying passengers (SOLAS Chapter V). In addition, national regulations may include other vessel types, such as per recent requirements by the European Union that all fishing vessels bigger than 15m must carry an AIS device [4]. Both VMS and AIS feature on-board transmitters linked to the vessel’s GPS to receive and transmit exact position in time and space on long-range radio frequencies to either coastal ground stations or satellites. In the case of AIS, data are also transmitted to other ships in the area that carry the device. VMS usually transmits in time intervals varying from one to several hours, satellite-based AIS (S-AIS) transmissions can be as frequent as every few seconds, enabling the monitoring of fine-scale vessel behavior and movement patterns. Several attempts have been made to use VMS and AIS data to understand fishing vessel behavior, for example by using simple presence/absence or vessel speed [5] [6]. While speed can be a useful indicator of vessel activity, operational speeds while fishing vary greatly for different fishing gear types such as trawls, longlines or nets. More sophisticated algorithms differentiating fishing from non-fishing activity for different fleets and gears are needed to properly capture and represent the characteristics of the various fishing methods, as stated previously by Natale et al. [4] We develop and present such algorithms here, then assess their accuracy in correctly identifying individual fishing events or ‘sets’ by comparing against expert-labeled data. Finally, we briefly chart potential applications in marine ecology, conservation, and fisheries management. 2 Methods 2.1 Data Sets This work is based on a database containing global S-AIS data obtained from AIS-enabled communication satellites since January 2011 until October 2015. Data were obtained under research licence from exactEarth (http://www.exactearth.com/products/exactais). A representation of three several-year tracks and examples of fishing activity patterns for trawling, longlining and purse seining is given in Fig 1. Individual tracks for (...truncated)