View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by St Andrews Research Repository Marine Pollution Bulletin 140 (2019) 17–29 Contents lists available at ScienceDirect Marine Pollution Bulletin journal homepage: www.elsevier.com/locate/marpolbul Review A review of unmanned vehicles for the detection and monitoring of marine T fauna ⁎ Ursula K. Verfussa, , Ana Sofia Anicetob,1, Danielle V. Harrisc, Douglas Gillespied, Sophie Fieldinge, Guillermo Jiménezf, Phil Johnstonf, Rachael R. Sinclaira, Agnar Sivertseng, Stian A. Solbøg, Rune Storvoldg, Martin Biuwb,2, Roy Wyattf a SMRU Consulting, New Technology Centre, North Haugh, St Andrews, Fife KY16 9SR, UK b Akvaplan-niva AS, Fram Centre, P.O. Box 6606, Langnes, 9296 Tromsø, Norway c Centre for Research into Ecological and Environmental Modelling, The Observatory, University of St Andrews, St Andrews, Fife KY16 9LZ, UK d Sea Mammal Research Unit, Scottish Oceans Institute, University of St Andrews, Fife KY16 8LB, UK e British Antarctic Survey, High Cross, Madingley Road, Cambridge CB3 0ET, UK f Seiche Ltd., Bradworthy Industrial Estate, Langdon Road, Bradworthy, Holsworthy, Devon EX22 7SF, UK g Norut - Northern Research Institute, Postboks 6434 Forskningsparken, 9294 Tromsø, Norway ARTICLE INFO ABSTRACT Keywords: Recent technology developments have turned present-day unmanned systems into realistic alternatives to tra- Unmanned vehicles ditional marine animal survey methods. Benefits include longer survey durations, improved mission safety, Marine animal monitoring mission repeatability, and reduced operational costs. We review the present status of unmanned vehicles suitable Underwater sound for marine animal monitoring conducted in relation to industrial offshore activities, highlighting which systems Environmental impact assessment are suitable for three main monitoring types: population, mitigation, and focal animal monitoring. We describe Offshore industry the technical requirements for each of these monitoring types and discuss the operational aspects. The selection of a specific sensor/platform combination depends critically on the target species and its behaviour. Thetech- nical specifications of unmanned platforms and sensors also need to be selected based on the surrounding conditions of a particular offshore project, such as the area of interest, the survey requirements and operational constraints. 1. Introduction monitoring’ to investigate fine-scale animal responses to anthropogenic sound. In recent decades, there has been increased awareness concerning Unmanned vehicles have the potential to greatly augment marine the potential impacts of underwater sound on marine animals, such as animal monitoring surveys. Some of the benefits of this technology auditory injury and/or behavioural changes (e.g., Gordon et al., 2003; compared to manned systems are improved mission safety, repeat- Ketten, 2014; Lucke et al., 2009; Pirotta et al., 2014). When conducting ability, and reduced operational costs. They also enable long-range industrial activities involving sound, such as pile driving or the use of operations beyond detection ranges of human observers. Unmanned seismic sources, regulators often prescribe monitoring before, during vehicles can be deployed in air (Unmanned Aerial Systems – UAS), at and/or after operations to assess and/or mitigate anthropogenic im- the sea surface (Autonomous Surface Vehicles – ASV) or in the water pacts on marine species. Three types of monitoring are typically used in column (Autonomous Underwater Vehicles – AUV). relation to industrial activities: (1) ‘population monitoring’ to assess Unmanned vehicles have quickly evolved over the past decade and animal abundance, density and/or distribution and changes therein; (2) are used in various studies, e.g., for gathering oceanographic and me- ‘mitigation monitoring’ to trigger mitigation actions upon animal pre- teorological data (e.g. Eriksen et al., 2001; Funaki and Hirasawa, 2008; sence near or within a potential impact area; and (3) ‘focal animal Leong et al., 2012; Meyer, 2016; Williams et al., 2010; Wynn et al., ⁎ Corresponding author. E-mail address: [email protected] (U.K. Verfuss). 1 ARCEx (Research Centre of Arctic Petroleum Exploration) UiT The Arctic University, Department of Geology, Dramsveien 201, Postboks 6050 Langnes, N-9037 Tromsø, Norway. 2 Institute of Marine Research, Tromsø Department, UK Sykehusveien 23, P. O. Box 6404, 9294 Tromsø, Norway. https://doi.org/10.1016/j.marpolbul.2019.01.009 Received 10 February 2018; Received in revised form 14 December 2018; Accepted 5 January 2019 0025-326X/ © 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/). U.K. Verfuss et al. Marine Pollution Bulletin 140 (2019) 17–29 2014), and for monitoring sea ice (e.g., Inoue et al., 2008) and wildlife photographs). In passive acoustic surveys where individuals cannot be (e.g., Forney et al., 2012; Hodgson et al., 2010; Koski et al., 2009b; identified acoustically (Section 5.2), SCR data are collected by asso- Lyons et al., 2006). Industry applications, such as for the Oil and Gas (O ciating the same acoustic detection across multiple hydrophones, where &G) industry, include subsea equipment inspections and leak detection the hydrophone locations are known (Stevenson et al., 2015). De- (Budiyono, 2009) and marine mammal monitoring around petroleum termination of the received sound levels and time-of-arrival of the operations (e.g., Koski et al., 2009a; Lyons et al., 2006). acoustic detections can improve the precision of SCR results (Stevenson We synthesize the findings of a review conducted by Verfuss et al. et al., 2015). (2015), highlighting which type of unmanned vehicle would be suitable When combining multiple systems or sensors into instrument arrays, for population, mitigation, and/or focal animal monitoring. Section 2 clock synchronisation will be essential to enable range estimation for describes the monitoring types and their demands on unmanned ve- distance sampling, and may also facilitate identifying the same acoustic hicles. Section 3 summarises the currently available unmanned plat- detection on multiple hydrophones for SCR. In addition to detection forms suitable for surveying marine mammals, sea turtles and fish. probability, other factors are likely to be required for abundance or Sections 4 and 5 present the data relay systems and sensor types, re- density estimation, e.g. group size, call production rate for acoustically spectively, which can be integrated with the platforms (forming an detected animals and surfacing rate for visually detected animals unmanned system). In Section 6 we evaluate which systems are cur- (Borchers et al., 2013; Marques et al., 2013a; Warren et al., 2017); such rently most suitable for each of the monitoring types and conclude in factors may require additional data collection from auxiliary fine scale Section 7 with a set of recommendations on future work. focal follows (Section 2.3). 2. Monitoring types 2.2. Mitigation monitoring 2.1. Population monitoring Mitigation monitoring is conducted to implement mitigation mea- sures upon the presence of certain marine animals within a pre-defined Population monitoring is used to estimate absolute population area around an anthropogenic sound source (mitigation zone) in order abundance or density, assess spatial and temporal patterns in the dis- to minimize the impact of sound on the animal. tribution of populations and investigate changes in density and dis- The size of the area that needs to be monitored (the monitoring tribution as a result of anthropogenic activities. zone) and which species are to be monitored for are generally specified Wildlife surveys for population monitoring typically take place by the responsible authority. The typical radius of a monitoring zone along planned survey transect lines or from static monitoring point ranges from 500 m (e.g., JNCC, 2017) to over 3000 m (e.g., DEWHA, transects, which are systematically placed across the study area. 2008) (see Verfuss et al., 2016 for further details). Detection probability Therefore, unmanned vehicles that follow transects with a minimum of of the target species should be high across the entire monitoring zone. operator oversight, by adhering to transect lines, or remaining sta- Accurate animal location or range determination are also desirable in tionary (or at least performing turns to stay in the same location) to order to avoid costly mitigation actions triggered by animals outside the create a monitoring point, are particularly suited to collecting data for mitigation zone. It is also essential that data processing can take place transect surveys. When selecting a platform, it is important to consider in near-real time, which either requires the transmission of raw data to the range of operating altitudes/depths, whether a system can follow a a competent human operator, or real time on-board processing (which pre-designed track (e.g. using pre-programmed coordinates, and/or may include the use of detectors and/or classifiers) and the transmis- manual piloting) with sufficient power to ensure the surveys arecom- sion of summary detection data, or a combination thereof (see Section pleted on time, and what environmental limitations may affect a sys- 4). It is generally necessary for a trained human operator to check de- tem's ability to remain on a survey path. tections before