- Email: [email protected]
Completing fishing monitoring with spaceborne Vessel Detection System (VDS) and Automatic Identification System (AIS) to assess illegal fishing in Indonesia
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
Contents lists available at ScienceDirect
Marine Pollution Bulletin journal homepage: www.elsevier.com/locate/marpolbul
Completing fishing monitoring with spaceborne Vessel Detection System (VDS) and Automatic Identification System (AIS) to assess illegal fishing in Indonesia Nicolas Longépéa,*, Guillaume Hajducha, Romy Ardiantob, Romain de Jouxa, Béatrice Nhunfata, Marza I. Marzukib,c, Ronan Fabletc, Indra Hermawanb, Olivier Germaina, Berny A. Subkib, Riza Farhanb, Ahmad Deni Muttaqinb, Philippe Gaspard a b c d
Space and Ground Segment, Collecte Localisation Satellites (CLS), Plouzané, France Agency for Marine and Fisheries Research and Development - MMAF, Jakarta, Indonesia Institut Mines-Télécom/Télécom Bretagne, CNRS UMR 6285 Lab-STICC, France Sustainable Management of Fisheries, Collecte Localisation Satellites (CLS), Ramonville Saint-Agne, France
A R T I C L E I N F O
A B S T R A C T
Keywords: Illegal fishing Synthetic Aperture Radar (SAR) Automatic Identification System (AIS) Vessel Monitoring System (VMS)
The Indonesian fisheries management system is now equipped with the state-of-the-art technologies to deter and combat Illegal, Unreported and Unregulated (IUU) fishing. Since October 2014, non-cooperative fishing vessels can be detected from spaceborne Vessel Detection System (VDS) based on high resolution radar imagery, which directly benefits to coordinated patrol vessels in operation context. This study attempts to monitor the amount of illegal fishing in the Arafura Sea based on this new source of information. It is analyzed together with Vessel Monitoring System (VMS) and satellite-based Automatic Identification System (Sat-AIS) data, taking into account their own particularities. From October 2014 to March 2015, i.e. just after the establishment of a new moratorium by the Indonesian authorities, the estimated share of fishing vessels not carrying VMS, thus being illegal, ranges from 42 to 47%. One year later in January 2016, this proportion decreases and ranges from 32 to 42%.
1. Introduction Illegal, Unreported, and Unregulated fishing activities, often referred as IUU, have become a global issue, threatening ocean ecosystems and sustainable fisheries (Agnew et al., 2009). IUU contributes to the overexploitation of resources, preventing recoveries and leading to collapses. With the perspective to feed an expected global population of 9 billion by 2050, fish already provides 16.7% of the global population's intake of animal protein and 6.5% of all protein consumed (FAO, 2014). This increasing demand will likely boost the amount of IUU activities in the world, and will be particularly severe in areas lacking effective conservation and management measures. With an Exclusive Economic Zone (EEZ) of about 7.9 million square kilometers, Indonesia has one of the largest maritime domains in the World (see Fig. 1). According to FAO (2012), there were about 2.6 million people engaged in fishing activities as fishers in 2010, with motorized marine fleet increasing by 11% from 348 425 fishing vessels in 2007 to 390 770 in 2009. Marine capture fisheries include demersal
*
and small-pelagic species fished on the continental shelf. e.g sardines, mackerels, scads, hairtail or snapper. Large pelagic species such as tuna (skipjack, bigeye and yellow fin) are rather caught in the mid and eastern part of the archipelago waters as well as in the Indonesian EEZ and on the high seas (Lehodey et al., 2017). As stated in Suhendar (2013), the fisheries resource in Indonesia has a potential of 6.4 million tons per year, with a current level of utilization of 5.81 million tons per year in 2012. Abundant commercial fish resources with weak patrol surveillance in a large EEZ is leading to Indonesia as one of the countries with the highest degree of IUU fishing activities in the world (Petrossian, 2015). Over-fishing, overcapacity and illegal fishing severely affect the sustainability of the fisheries. They cause catastrophic economic, social and environmental losses. More than half of the income from fishing that should benefit to Indonesians is misappropriated by illegal fishing, and this action also weakens Indonesia's sovereignty. The Arafura Seas are often cited as one of the most critical areas for IUU activities (Nurhakim et al., 2008). The Arafura Sea is also the most important
Corresponding author. E-mail address: [email protected] (N. Longépé).
http://dx.doi.org/10.1016/j.marpolbul.2017.10.016 Received 22 December 2016; Received in revised form 29 September 2017; Accepted 6 October 2017 0025-326X/ © 2017 Elsevier Ltd. All rights reserved.
Please cite this article as: Longepe, N., Marine Pollution Bulletin (2017), http://dx.doi.org/10.1016/j.marpolbul.2017.10.016
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
Fig. 1. Map of Indonesia with INDESO satellite receiving system and its visibility circle in orange, coverage of the acquired SAR images used in this study (in red) together with their associated VMS vessels (blue dots). The Indonesia EEZ from Claus et al. (2017) is indicated by the black line. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
fishing ground for demersal fisheries. Fish trawls, shrimp trawls and bottom long lines are mainly used (Nurhakim et al., 2008). A detailed description of IUU fishing in Indonesia and more specifically in the Arafura Sea is provided in Resosudarmo et al. (2009). The Monitoring, Controlling and Surveillance (MCS) of fishing has been undertaken with the objective in fine to manage and exploit fish resources in a responsible manner (FAO, 1995). The Indonesian Ministry of Marine Affairs and Fisheries (MMAF) has lately considered this issue to be of national importance. It is taking action and introducing firm measures against the perpetrators of illegal fishing in the interest of Indonesia's territorial sovereignty. The MMAF selected the Argos technology in 2002 for its Vessel Monitoring System (VMS) to regulate fisheries. At that time, 1500 units Argos MARGE transmitters were fitted to fishing vessels, connected to a Fishing Monitoring Center (FMC) in Jakarta. The first concepts of integrating spaceborne high-resolution radar satellite data (so called Synthetic Aperture Radar - SAR) with VMS for fisheries monitoring were demonstrated in the European waters by Kourti et al. (2001). Since 2004, a pioneering service based on SAR data is operational without interruption at the Kerguelen Island, a French territory in the southern Indian ocean. By creating the INDESO (Infrastructure Development of Space Oceanography) project in 2012, Indonesia is taking a significant step forward to fight against illegal fishing activities. The implementation of an operational IUU Fishing application within the INDESO project strongly relies on the INDESO satellite receiving system located at Perencak, Bali, Indonesia. The center has been in operation since October 2014. It can acquire, process and analyze high-resolution SAR imagery enabling to detect non-cooperative fishing vessels via a Vessel Detection System (VDS). In these radar images, metallic vessels appears as bright echoes that can be detected via image processing. In addition, the use of satellite Automatic Identification System (satAIS) tracking data completes the solution by providing maritime awareness for most of vessels including larger vessels such as tankers or cargo. It is of particular interest in areas with mixed traffic where the VDS/VMS is not sufficient to identify IUU activities. The integration of such data was supported by a set of R & D programs such as Greidanus (2007) and EU (2013). In this paper, the benefit of SAR-based VDS in completing VMS over the Arafura Seas is presented. This study does not consider unreported and unregulated fishing activities, but solely illegal fishing. In particular, an estimate of the amount of illegal fishing activities via a systematic correlation between VMS-tracked fishing vessels and SAR-based echoes is provided. By integrating Sat-AIS data in the analysis, and thus
removing the SAR echoes corresponding to other type of vessels (e.g. cargo, tankers), non-correlated SAR echoes are potentially illegal fishing vessels. Section 2 describes the newly operational INDESO system. In particular, the interest of Sat-AIS and SAR-based VDS are highlighted. The joint analysis of AIS, SAR-based VDS reports and VMS data is then performed via a systematic matching process. An attempt to provide the share of illegal fishing activities is provided in Section 3. Section 4 concludes this study. 2. Description of INDESO system and subsystems 2.1. VMS data The Indonesian MCS strategy strongly relies on data analysis performed by the FMC where VMS data have been originally handled. It is actually recognized as a cornerstone for any fisheries MCS system. VMS is used by national regulatory authorities, and is not public such that sensitive positional information about fishing areas are not available to other fishing vessels. In Indonesia, it has been activated since 2003 when national authorities noticed a significant decrease in fish stocks and in the mean size of individual species. A description of the VMS system in Indonesia is provided in Suhendar (2013). To date, VMS messages are sampled every 60 min (24 positions per day plus eventually 12 complementary upon specific programming) with information on vessel speed and route, eventually log-book, assistance messages or other information such as estimated arrival time. Several studies have also demonstrated the relevance of VMS trajectories data to monitor fishing activities such as presence of a vessel within exclusion zones, gear types and fishing methods (Bertrand et al., 2007; Bez et al., 2011; Joo et al., 2011; Marzuki et al., 2015). In the following, VMS data made available from October 2014 to March 2015, and in January 2016 are used. 2.2. Rationale The new concept of IUU monitoring is based on the joint use of VMS and VDS based on SAR imagery. All vessels (above a reasonable size) can be detected by images provided by SAR sensors over a monitored area, independently on the cloud cover and light condition (day/night). Cooperative fishing vessels transmit VMS signals while non-cooperative vessels are spotted by radar imagery with no corresponding VMS positions. Non-cooperative vessels are more likely to be poachers. Based on the joint analysis of VMS and VDS data, law enforcement stakeholders can plan a surveillance operation and send a patrol boat to 2
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
Time responsiveness is a key element in maritime surveillance, so that maritime patrols can be associated with satellite radar detections to optimize IUU surveillance operations. For the automatic processes, radar images are produced, indexed by the system, and automatic ship detection and characterization are made available in about 10 min after sensing. The results are then available for further analysis and manual editing by trained image experts in the INDESO operation room. As soon as the VDS report is available (about 10–20 min later), a dedicated GIS platform automatically ingests the ship detection report and performs the automatic correlation with VMS data that have been previously ingested. For the association process, the VMS data are interpolated between position reports at times either side of the image collection. The association process can lead to three possible situations:
inspect the suspected target. This is an effective solution to arrest violating boats without any license or not equipped with VMS. The high cost of satellite radar data is a potential issue for sustained operational. The Agency for Marine and Fisheries Research and Development (AMFRD) of MMAF has recently implemented its own infrastructure operated since October 2014 via the INDESO project. The service provided by the INDESO infrastructure uses the vessel detection reports generated in Near-Real Time (NRT - typically in less than 30 min after each satellite pass) at the newly built satellite receiving station located in Perancak, Bali, Indonesia. Vessel detection reports' are then merged with the VMS data into a dedicated GIS system to allow trained operators to perform fishing detection using both ad hoc matching algorithms and contextual information. Whereas fishing vessels in the European Union have been gradually equipped with AIS transmitters (since May 2014, all vessels over 16 m), no such regulation exists for Indonesian waters. However, as stated by the International Maritime Organization, AIS transmitters shall be fitted aboard all ships of 300 gross tonnage and upwards engaged on international voyages, cargo ships of 500 gross tonnage and upwards not engaged on international voyages and all passenger ships irrespective of size. AIS data may therefore be of high interest to track some foreign fishing vessels, and correlate VDS report with non-fishing vessels. At this stage, it should be noted that Long Range Identification Tracking (LRIT) system is not used our study. This system which apply to certain type of ships engaged on international voyages report the positions of vessels to their flag administration only.
• The detected SAR echo is associated with one VMS vessel, which is the nominal case for fishing vessel; • The detected SAR echo is associated with one AIS vessel, which is •
The automatic vessel detection is based on state-of-the-art algorithms described in Hajduch and Kerbaol (2006) with detailed descriptions of assessment strategy in Longépé et al. (2013) and Pelich et al. (2015). The incidence angle strongly impacts the detection capabilities, the larger the better. However, the number of false alarms may increase as well. From our experience, we estimate that, after an analysis by trained operators, less than 10% of detected echoes remain false alarms for Fine Wide imagery sensed in between 31° to 39°. This ratio increases to about 15% for incidence angle in between 39° to 45°. This remains a challenge for trained operators, especially in the context of NRT services with limited amount of time. Breaking waves, radar-processing ambiguities artifacts which are sometimes tricky to discriminate (especially range ambiguities from land more than hundreds kilometers away - especially true for Indonesian archipelagos) and radar image quality may impact the quality of Vessel Detection Reports. For detailed explanations of these artifacts, the reader may refer to Crisp (2004). To assess the detection rate of fishing vessels with the Wide Fine acquisitions, a previous study (Pelich et al., 2015) is used as a proxy. The detection rate depends on many parameters which are beyond the scope of this study (inter alia sea state & wind speed, incidence angle, the number of looks, the polarization, the pixel spacing/sampling, the radar resolution, the radar cross section of vessel, the frequency of electromagnetic waves...). Here, the assessment is simplified assuming the detection rate depends on the pixel spacing (proxy of radar resolution by a factor 2) and the incidence angle only. From Pelich et al. (2015), the rate of good detection for other sensing modes of the Radarsat-2 satellite is modified considering 1) the new ratio of vessel size w.r.t. pixel spacing and 2) a re-normalization of the impact of the incidence angle via the Figs. 4 and 5 of Pelich et al. (2015) and the ones of our dataset. The estimated rates of good detection with respect to vessel length are provided in Fig. 2.
2.3. VDS with high resolution radar satellites The VDS is based on the capacity to receive and process high resolution satellite radar imagery in NRT via a dedicated satellite receiving station. The satellite receiving antenna (5.4 m diameter) and its processor terminal can receive radar images from two types of satellites, namely the Canadian Radarsat-2 and the Italian CosmoSkyMed satellites. Monitoring requirements are translated into satellite acquisition plans and sent to the satellite operating agency. This task is performed by the INDESO operations on a quarterly basis. In case of surveillance operations at sea, it is of the utmost importance that the route of the patrol vessels is guided by the satellite acquisition plan and not the reverse since there is no way to change the satellite orbits. From October 2014 to February 2017, 896 Radarsat-2 and 130 CosmoSkyMed images were received, processed and analyzed in NRT at INDESO station and 960 ship detection reports were delivered. For Radarsat-2 satellites, two modes (so called Wide and Wide Fine) are mostly acquired as they offer the capabilities to detect small vessels over relatively large area. As of end 2016, the current rate of SAR image acquisitions is about 28 WideFine Radarsat-2 images per month. In this study, SAR data acquired over the Arafura Sea concomitantly with available VMS data are used, i.e. from October 2014 to March 2015, and in January 2016. For the first and second periods of analysis, 139 and 9 images are used, respectively. 120 among these 148 images have been acquired with incidence angle ranging from 31° to 39°. The main characteristics of these two modes are provided in Table 1. Table 1 Characteristics and availability of Radarsat-2 imagery in the INDESO database with available VMS data. The abbreviations rg. and az. stand for the range and azimuth directions, respectively. Beam mode
Wide Wide fine
Nb. images
Coverage
Pixel spacing
Radar resolution
2014–2015/ 2016 11/0 128/9
rg. × az. (km)
rg. × az. (m)
rg. × az. (m)
150 × 150 150 × 170
12.5 × 12.5 6.25 × 6.25
40.0–19.2 × 24.7 15.2–8.2 × 7.7
the nominal case for non-fishing vessel. For instance, the sat-AIS data over the Arafura Sea contain a very limited number of fishing vessels, but mostly tankers or cargoes; The detected SAR echo is associated with neither AIS nor VMS vessels. Depending on whether its length estimated from the SAR image is below 50 m, it may be considered as a suspected fishing vessel.
2.4. Sat-AIS The AIS uses the maritime Very High Frequency (VHF) band for the transmission and reception of data. It was first implemented as a technology to avoid collisions among large vessels at sea and exchange information such as identification, position, course and speed. AIS messages are sent from all merchant ships above 300 gross tons engaged in international voyages, ships above 500 GT in domestic 3
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
terrestrial AIS receivers are installed onshore. For class-B vessel, this technology may be less effective as the transmit power is rather low and close to the noise floor of the system. A comprehensive overview of AIS technology is provided in Carson Jackson (2012). As opposed to VMS, AIS data are aimed at being publicly available as a commercial product. The introduction of error, the falsification of the AIS messages or spoofing attacks are increasingly found:
• Non-intentional errors can be made by manually entering data. It •
Fig. 2. Estimated good detection rate of vessels from Radarsat-2 fine wide mode, depending upon the range of incidence angles of the acquired image.
•
voyages, and all passenger vessels. Most of large vessels are equipped with a so-called Class-A transceiver (transmit and receive) with 12.5 W transmit power and the capability to reserve transmit slot via 2 receivers in continuous operation. Class-A AIS transponder broadcasts positional messages every 2 to 10 s while underway, and every 3 min while at anchor. For leisure vessel and other smaller vessel, lighter solution - so called class-B - exists with less transmit power (2 W) and less sophisticated solution for transmit slot identification. The broadcast rate is nominally every 30 s or 3 min depending on vessel speed. Beyond position coordinates, AIS data contains Maritime Mobile Service Identities (MMSI) information including Maritime Identification Digits which are a direct identifier for vessel's flag. More specifically, the first digit indicates the region of origin with “3”, “4”, and “6”, the “North and Central America and Caribbean”, “Asia”, and “Africa”, respectively. While AIS is originally a “line of sight” technology with a coverage limit of about 20 nautical miles around the spherical Earth with vesselto-vessel or vessel-to-terrestrial transceivers, satellite-based AIS receivers is now widely used to ensure data reception even when no
may concern destination, vessel type but also MMSI information, 2% of cases according to Iphar et al. (2015). The falsification in the messages is clearly an intentional action aiming at hindering any potential surveillance systems. The most common falsification is to turn off AIS transmitters, which may hide illicit activities. It is important to note that turning off AIS is not, in and of itself, evidence of illicit activity. In some specific areas, this may be used to protect from pirates or to protect commercially sensitive information such as fishing grounds. The spoofing of messages is a voluntary action “from an actor which is external to the vessel, in order to mislead both the crew on board and the outside world on the behavior of a vessel.” (Iphar et al., 2015).
The first two items severely affect the potential of AIS to provide a comprehensive maritime awareness, not to mention the time latency and time revisit. New AIS-receiver satellite constellation has recently started to unroll from beginning of 2017 with claimed latencies and revisit of few minutes. As an illustration of the above behaviors, sat-AIS data from the Arafura Sea are analyzed. More specifically, the period before and after the establishment of the moratorium by MMAF's Regulation 56/2014 is studied. Since November 2014, foreign fishing vessels over 30 GT must renew their fishing license annually, and intensive investigation on suspected ex- foreign vessel has been conducted. Whereas the number of AIS messages transmitted by all the other types of vessel fluctuate with no real trend from November 2014, the number of vessels identified as “Fishing Vessel” drastically decreased from October to December 2014 (see blue curve in Fig. 3b). Fig. 3. The averaged number of vessels transmitting on AIS, and their declared type in the Arafura Sea: large vessels (a) and smaller vessels (b). The dash vertical line indicates the establishment of the MMAF moratorium for ex-foreign fishing vessel. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
4
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
“Leisure ship” while the pictures associated to their MMSI indicate fishing vessel. As given by VMS license, most of the fishing vessels are smaller than 30 m in the area of interest, few of them being however larger up to 50 m. In the area of interest, few other types of vessels are smaller than 50 m (see Fig. 3b). To decipher fishing from non-fishing vessels from SAR imagery, the estimated length of the detected echoes can thus be used as a decision rule, based on the good reliability of this estimator. To this end, the sizes of SAR-based echoes are compared with actual vessel length from VMS fishing vessels via a VMS-SAR matchups database. The results are shown in Fig. 7. Only 11 vessels among 515 have an estimated length over 50 m (which represent only 2.1% of the cases). This amount is considered as negligible in the following calculation. As a consequence, all SAR-based vessels whose estimated length is below 50 m are assumed to be fishing vessels. Considering all the above assumptions, the ratio of illegal fishing Rillegal can be estimated based on the assumption that legal fishing vessels carry VMS transceiver (see Fig. 6). With Nillegal being the number of illegal fishing vessels among all Nfish fishing vessels, we have from October 2014 to April 2015:
Fig. 4. Suspected vessels in the Arafura Sea via an analysis of MMSI. The dash vertical line indicates the establishment of the MMAF moratorium for ex-foreign fishing vessel.
Via an analysis of fishing vessels' MMSI for all AIS messages received from July 2014 to April 2016, most of fishing vessels had Asian flags (31 over 38). Only two had MMSI indicating North and Central America and Caribbean flag with over 1500 GT and 70 m length. One was registered from Western Africa. Last but not least, four of them had “fake” MMSI not following the rules of the International Telecommunication Union (e.g. 100000000, 110000000, 123000000, 123456789). As illustrated in Fig. 4, all vessels with MMSI starting with “3”, “4” or “6” actually left the Arafura Sea due to the moratorium. At this stage, two assumptions can be raised 1) these vessels went fishing in other seas or 2) they adapted their AIS transceiver filling erroneous MMSI or vessel type information. In any cases, the moratorium has clearly induced a change of behavior for foreign fishing fleets. The Sat-AIS data cannot provide on its own the solution for an efficient monitoring system for fishing vessels. In the following, it is essentially used to identify large vessels such as cargo or tankers.
⎧ Nillegal ⎨ Nfish ⎩
= 2 + 2 + 1 + (671 − 67) = 609 = 736 + (671 − 67) + 2 + 2 + 1 = 1345
A a result, approximately 45.3% of fishing vessels in the area are illegal (in the sense that they are not compliant with VMS) during the first period of analysis (from Oct. to April 2015). In January 2016 with the few data available, this amount is estimated to be 36.1%. 3.2. Refined estimates The rough estimate does not take into account bias and errors due to the proposed methodologies. The potential issues are numerous, linked to the SAR-based detection, to the temporal sampling of AIS and/or VMS, the methodologies to interpolate AIS/VMS tracks and then to correlate with SAR-based echoes. In the following, some issues are considered, discussed, and their impact on the above estimate evaluated. As indicated by Fig. 2, the probability of SAR-based detection pd for vessels with length above 15 m but below 50 m ranges from 80 to 100%. Considering that the fishing vessels that were not detected by the SAR detector are partly legal and illegal with exactly the same ratio as of all the other fishing vessels, we thus have
3. Towards the assessment of illegal fishing In this section, an assessment of the proportion of illegal fishing activities is provided. To do so, the cross-analysis of SAR-derived vessel detections and VMS/AIS tracks is used from the 148 Radarsat-2 images mentioned in Section 2.3. AIS and VMS tracks are interpolated at the time of each SAR acquisition time, and interpolated echoes are then matched with corresponding SAR echoes following distance criteria (see Figs. 5 and 6). A post-processing analysis is carefully operated via GIS software to remove any erroneous data, e.g. close to shoreline or corresponding to mooring areas.
⎧ Nillegal = 609 + pillegal (NfishSAR/ pd − NfishSAR)) ⎨ Nfish = 1345 + (NfishSAR/ pd − NfishSAR) ⎩ with NfishSAR = (671 − 67) + 538 + 10 + 4 = 1156 being the number of fishing vessels as detected by the SAR detector. After solving this equation, the probability of detection pd does not affect the computation of illegal fishing rate. In addition, the false alarm rate for the SAR-based vessel detector should be considered as mentioned in Section 2.3. Here, a false alarm rate RFA ranging from 5 to 15% for all vessels below 50 m is assumed. Above 50 m, no false alarm is considered. It is assumed that false alarms only impact SAR targets non-correlated with VMS echoes, SAR targets associated to VMS being “true” vessels. A set of falsely (671 − 67)RFA echoes should then be removed from the statistics, the ratio of illegal fishing vessel becomes
3.1. Evaluation principles Before any attempt to assess the share of illegal over legal fishing from this data, a set of assumptions is considered. The VMS equipped vessels are fishing vessels and all of them are fishing legally. This hypothesis is an overestimation of the fleet of legal fishing vessels as vessel can transmit their position using VMS technology while having an invalid or expired fishing license. In addition, being VMS compliant does not preclude from having legal fishing activities. At this stage, it should be recall that this study does not consider unreported and unregulated fishing activities. With respect to AIS, the vessels reporting their AIS types as “fishing” are actually fishing vessels. The “type” field of AIS message does not necessarily reflect the real type of the vessel. However, the vessels fishing legally do not have reason to spoof this field whereas the vessels fishing illegally may have reason to spoof it. This can be refined by manual inspection of the database of each unique vessel. For instance, two potential fishing vessels are spotted as they report their type as
Rillegal =
609 − 604RFA 1345 − 604RFA
Vessels equipped with VMS transmit their positions regularly every 60 min. While performing the matching process, the time lag between one of the two VMS messages and the SAR echoes should be thus 30 min maximum. Erroneous VMS data with sampling exceeding 60 min are discarded. In addition, the maximum distance between SAR 5
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
Fig. 5. Map of SAR-VMS-AIS matchups from INDESO database (from October 2014 to April 2015, and in January 2016). Red square: VMS unmatched, green square: VMS-SAR matchups, blue circle: AIS unmatched, pink circle: AIS-SAR matchups, and black star: SAR unmatched. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
and interpolated VMS positions is heightened as 20 km to ensure the highest number of matchups. However, a fraction of them might be considered as erroneous, thus adding a certain number of SAR-based targets as potential illegal fishing vessel. This mismatch rate Rmis may range up to 20% out of the 538 matchups. From October 2014 to April 2015, the ratio of illegal fishing vessel thus depends on the SAR false alarms rate and the rate of mismatch between VMS and SAR:
Rillegal =
609 − 604RFA + 538Rmis Nfish/ Nvessel 1345 − 604RFA + 538Rmis Nfish/ Nvessel
where Nvessel is the number of all vessels in the areas observed by the SAR imageries (incl. tankers...). It is equal to Nvessel = 736 + 157 + 46 + 671 = 1610. The above equation and the one corresponding to January 2016 are illustrated by Fig. 8. From October 2014 to April 2015, the estimated share of illegal fishing activities with respect to the total fishing activities in the Arafura Sea ranges from about 42 to 47% with RFA in the [5–15]% interval. In January 2016, it ranges from about 32 to 42%. At this stage, it is recall
Fig. 7. Assessment of vessel length estimation based on 515 SAR-VMS matchups and license information from MMAF - among the 738 matchups, some vessel lengths were missing in the VMS MMAF license information.
Fig. 6. Overview of SAR-VMS-AIS matchups from INDESO database: from October 2014 to April 2015, corresponding to 139 analyzed SAR images (left), and in January 2016 corresponding to 9 analyzed SAR images (right).
6
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
Acknowledgments This study has been funded by the INDESO project and the Indonesian Ministry of Fisheries and Marine Affairs. In addition, some technical processes (sat-AIS data manipulation, SAR/VDS coupling) were partly co-funded by ESA/ESTEC via the AMTRAC project and ARTES 20 program. We would like to thank the Editor and the two anonymous reviewers for their fruitful comments and suggestions. References Agnew, J., Pearce, J., Ganapathiraju, P., Peatman, T., Watson, R., Beddington, J., Pitcher, T., 2009, 02. Estimating the worldwide extent of illegal fishing. PLoS One 4, 1–8. Bertrand, S., Bertrand, A., Carrasco, R.G., Gerlotto, F., 2007. Scale invariant movements of fishermen: the same foraging strategy as natural predators. Ecol. Appl. 17, 331–337. Bez, N., Walker, E., Gaertner, D., Rivoirard, J., Gaspar, P., 2011. Fishing activity of tuna purse seiners estimated from vessel monitoring system (VMS) data. Can. J. Fish. Aquat. Sci. 68, 1998–2010. Carson Jackson, J., 2012. Satellite AIS - developing technology or existing capability? J. Navig. 6, 303–321. Claus, S., De Hauwere, N., Vanhoorne, B., Dias, S., Oset García, P., Hernandez, F., Mees, J., 2017. Database marine regions. Technology Fact Sheets. Crisp, D., 2004. The state-of-the-art in ship detection in Synthetic Aperture Radar imagery. In: DSTO. EU, 2013. Final report summary - DOLPHIN (development of pre-operational services for highly innovative maritime surveillance capabilities). In: Technical Report. Project ID: 263079. FAO, 1995. Code of conduct for responsible fisheries (CCRF). In: Technical Report. Food and Agriculture Organisation of the United Nations, ROME. FAO, 2012. The state of world fisheries and aquaculture. In: Technical Report. FAO Fisheries and Aquaculture Department. FAO, 2014. The State of World Fisheries and Aquaculture: Opportunities and Challenges. 978-92-5-108275-1 (E-ISBN 978-92-5-108276-8, Rome. Greidanus, H., 2007. DECLIMS: detection, classification and identification of marine traffic from space: final report. In: Technical Report. EC-JRC. Hajduch, G., Kerbaol, V., 2006. Ships detection on envisat asar data: results, limitations and perspectives. In: SEASAR 2006, ESA/ESRIN, Frascati. Husein, S.Y., 2015. Strategy on combating IUU fishing and post moratorium policies plan. In: FishCRIME, Cape Town, South Africa. Iphar, C., Napoli, A., Ray, C., 2015, October, October. Detection of false AIS messages for the improvement of maritime situational awareness. In: Oceans2015, Washington, DC, United States. Joo, R., Bertrand, S., Chaigneau, A., iquen, M., 2011. Optimization of an artificial neural network for identification of fishing event positions from vessel monitoring system data. Ecol. Model. 222, 1048–1059. Kourti, N., Shepherd, I., Schwartz, G., Pavlakis, P., 2001. Integrating spaceborne SAR imagery into operational systems for fisheries monitoring. Can. J. Remote. Sens. 4, 291–305. Lehodey, P., Senina, I., Wibawa, T., Titaud, O., Calmettes, B., Conchon, A., Tranchant, B., Gaspar, P., 2017. Operational modelling of bigeye tuna (Thunnus obesus) spatial dynamics in the Indonesian region. Mar. Pollut. Bull (in press). Longépé, N., Hajduch, G., Pelich, R., Habonneau, J., Lebras, J., 2013. SAR-based ship monitoring: advanced methodologies with medium resolution images (from WSM ASAR to EWS/IWS S1 mission). In: ESA Living Planet Symposium. Marzuki, M.I., Gaspar, P., Garello, R., Kerbaol, V., Fablet, R., 2015. Fishing gear recognition from VMS data to identify illegal fishing activities in Indonesia. In: OCEANS 2015 - Genova. Nurhakim, S., Nikijuluw, V., Badrudin, M., Pitcher, T., Wagey, G., 2008. A study of illegal, unreported and unregulated (IUU) fishing in the Arafura Sea, Indonesia. In: Technical Report. FAO, Rome. Pelich, R., Longépé, N., Mercier, G., Hajduch, G., Garello, R., 2015. AIS-based evaluation of target detectors and SAR sensors characteristics for maritime surveillance. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 8, 3892–3901. Petrossian, G., 2015. Preventing illegal, unreported and unregulated (IUU) fishing: a situational approach. Biol. Conserv. 189, 3948. Purwanto, 2013, September. Fishing fleet productivity and potential production of shrimp fishery in the Arafura Sea. J. Lit. Perikan. Ind. 19, 147–155. Resosudarmo, B., Napitupulu, L., Campbell, D., 2009. Working With Nature Against Poverty: Development, Resources and the Environment in Eastern Indonesia: Illegal Fishing in the Arafura Sea. ISEASYusof Ishak Institute. Sodik, D.M., 2007. Combating illegal, unreported and unregulated fishing in Indonesian waters: the need for fisheries legislative reform. University of Wollongong Ph.D. thesis. Suhendar, M., 2013. Comparison of vessel monitoring system (VMS) between iceland and indonesia [final project]. In: Technical Report. Ministry of Marine Affairs and Fisheries.
Fig. 8. Estimated rate of illegal fishing activities in the Arafura Sea based on VMS-SARAIS matchups for the two analyzed periods: from October 2014 to April 2015 with solid line, and January 2016 with dashed line.
this estimate is based on the major assumption that all VMS equipped vessels are legal with no consideration on their associated license and their activity. The joint analysis of the VMS/SAR/AIS correlation and VMS-related license data is left to another study. Some vessels may breach the conditions of their licenses (Sodik, 2007). From data extracted in 2011, Purwanto (2013) stated the amount of unlicensed shrimp trawl vessels is about 63%. However, the number of shrimp trawlers compared to fish trawlers is low (in 2005, 301 compared to 842 (Purwanto, 2013)), not to mention bottom long lines vessels. The estimated amount of illegal fishing activities from this study should not be compared to this amount, the period of analysis being different as well. With a new moratorium decided by the Indonesian government from November 2014, and the operational phase of the INDESO system starting from October 2014, illegal fishing activities should be decreasing, as it is observed from this study. 4. Conclusion In this study, the benefit of a complete MCS system is demonstrated. Especially, the added-value of AIS and SAR-based VDS complementary to that provided by VMS is exposed. The INDESO project provides the Indonesian MCS with the capacity to acquire and process SAR imagery. With its NRT capability, SAR-based VDS combined with VMS data and patrol vessels is a solution to prevent, deter and eliminate IUU fishing. In our analysis, the joint analysis of VMS-AIS and VDS report has been carried out, leading to an estimated amount of illegal fishing in the Arafura Sea based on the assumption that legal fishing vessels carry VMS systems. The considered period is mainly from October 2014 to March 2015, i.e just after the establishment of new moratorium by the MMAF in November 2014. It has been estimated that non-cooperative fishing vessels represent between 42 and 47% of the total number of fishing vessels in the Arafura Sea at that period of time. Even though the number of data available in January 2016 is limited, the share of illegal fishing activities seems lower almost a year later (up to −15% reaching potentially 32%). As stated in Husein (2015), a task force on Prevention and Eradication of IUU established by the MMAF leads to the compliance audit of 1132 ex-foreign vessels from March to October 2015. This audit plus a series of law enforcement actions supported by the INDESO infrastructure and relayed by mass media during 2015 surely impacted the amount of illegal but also unreported and unregulated fishing. The expected decrease should be further consolidated and monitored continuously afterwards. This research provides a baseline for future monitoring.
7
Contents lists available at ScienceDirect
Marine Pollution Bulletin journal homepage: www.elsevier.com/locate/marpolbul
Completing fishing monitoring with spaceborne Vessel Detection System (VDS) and Automatic Identification System (AIS) to assess illegal fishing in Indonesia Nicolas Longépéa,*, Guillaume Hajducha, Romy Ardiantob, Romain de Jouxa, Béatrice Nhunfata, Marza I. Marzukib,c, Ronan Fabletc, Indra Hermawanb, Olivier Germaina, Berny A. Subkib, Riza Farhanb, Ahmad Deni Muttaqinb, Philippe Gaspard a b c d
Space and Ground Segment, Collecte Localisation Satellites (CLS), Plouzané, France Agency for Marine and Fisheries Research and Development - MMAF, Jakarta, Indonesia Institut Mines-Télécom/Télécom Bretagne, CNRS UMR 6285 Lab-STICC, France Sustainable Management of Fisheries, Collecte Localisation Satellites (CLS), Ramonville Saint-Agne, France
A R T I C L E I N F O
A B S T R A C T
Keywords: Illegal fishing Synthetic Aperture Radar (SAR) Automatic Identification System (AIS) Vessel Monitoring System (VMS)
The Indonesian fisheries management system is now equipped with the state-of-the-art technologies to deter and combat Illegal, Unreported and Unregulated (IUU) fishing. Since October 2014, non-cooperative fishing vessels can be detected from spaceborne Vessel Detection System (VDS) based on high resolution radar imagery, which directly benefits to coordinated patrol vessels in operation context. This study attempts to monitor the amount of illegal fishing in the Arafura Sea based on this new source of information. It is analyzed together with Vessel Monitoring System (VMS) and satellite-based Automatic Identification System (Sat-AIS) data, taking into account their own particularities. From October 2014 to March 2015, i.e. just after the establishment of a new moratorium by the Indonesian authorities, the estimated share of fishing vessels not carrying VMS, thus being illegal, ranges from 42 to 47%. One year later in January 2016, this proportion decreases and ranges from 32 to 42%.
1. Introduction Illegal, Unreported, and Unregulated fishing activities, often referred as IUU, have become a global issue, threatening ocean ecosystems and sustainable fisheries (Agnew et al., 2009). IUU contributes to the overexploitation of resources, preventing recoveries and leading to collapses. With the perspective to feed an expected global population of 9 billion by 2050, fish already provides 16.7% of the global population's intake of animal protein and 6.5% of all protein consumed (FAO, 2014). This increasing demand will likely boost the amount of IUU activities in the world, and will be particularly severe in areas lacking effective conservation and management measures. With an Exclusive Economic Zone (EEZ) of about 7.9 million square kilometers, Indonesia has one of the largest maritime domains in the World (see Fig. 1). According to FAO (2012), there were about 2.6 million people engaged in fishing activities as fishers in 2010, with motorized marine fleet increasing by 11% from 348 425 fishing vessels in 2007 to 390 770 in 2009. Marine capture fisheries include demersal
*
and small-pelagic species fished on the continental shelf. e.g sardines, mackerels, scads, hairtail or snapper. Large pelagic species such as tuna (skipjack, bigeye and yellow fin) are rather caught in the mid and eastern part of the archipelago waters as well as in the Indonesian EEZ and on the high seas (Lehodey et al., 2017). As stated in Suhendar (2013), the fisheries resource in Indonesia has a potential of 6.4 million tons per year, with a current level of utilization of 5.81 million tons per year in 2012. Abundant commercial fish resources with weak patrol surveillance in a large EEZ is leading to Indonesia as one of the countries with the highest degree of IUU fishing activities in the world (Petrossian, 2015). Over-fishing, overcapacity and illegal fishing severely affect the sustainability of the fisheries. They cause catastrophic economic, social and environmental losses. More than half of the income from fishing that should benefit to Indonesians is misappropriated by illegal fishing, and this action also weakens Indonesia's sovereignty. The Arafura Seas are often cited as one of the most critical areas for IUU activities (Nurhakim et al., 2008). The Arafura Sea is also the most important
Corresponding author. E-mail address: [email protected] (N. Longépé).
http://dx.doi.org/10.1016/j.marpolbul.2017.10.016 Received 22 December 2016; Received in revised form 29 September 2017; Accepted 6 October 2017 0025-326X/ © 2017 Elsevier Ltd. All rights reserved.
Please cite this article as: Longepe, N., Marine Pollution Bulletin (2017), http://dx.doi.org/10.1016/j.marpolbul.2017.10.016
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
Fig. 1. Map of Indonesia with INDESO satellite receiving system and its visibility circle in orange, coverage of the acquired SAR images used in this study (in red) together with their associated VMS vessels (blue dots). The Indonesia EEZ from Claus et al. (2017) is indicated by the black line. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
fishing ground for demersal fisheries. Fish trawls, shrimp trawls and bottom long lines are mainly used (Nurhakim et al., 2008). A detailed description of IUU fishing in Indonesia and more specifically in the Arafura Sea is provided in Resosudarmo et al. (2009). The Monitoring, Controlling and Surveillance (MCS) of fishing has been undertaken with the objective in fine to manage and exploit fish resources in a responsible manner (FAO, 1995). The Indonesian Ministry of Marine Affairs and Fisheries (MMAF) has lately considered this issue to be of national importance. It is taking action and introducing firm measures against the perpetrators of illegal fishing in the interest of Indonesia's territorial sovereignty. The MMAF selected the Argos technology in 2002 for its Vessel Monitoring System (VMS) to regulate fisheries. At that time, 1500 units Argos MARGE transmitters were fitted to fishing vessels, connected to a Fishing Monitoring Center (FMC) in Jakarta. The first concepts of integrating spaceborne high-resolution radar satellite data (so called Synthetic Aperture Radar - SAR) with VMS for fisheries monitoring were demonstrated in the European waters by Kourti et al. (2001). Since 2004, a pioneering service based on SAR data is operational without interruption at the Kerguelen Island, a French territory in the southern Indian ocean. By creating the INDESO (Infrastructure Development of Space Oceanography) project in 2012, Indonesia is taking a significant step forward to fight against illegal fishing activities. The implementation of an operational IUU Fishing application within the INDESO project strongly relies on the INDESO satellite receiving system located at Perencak, Bali, Indonesia. The center has been in operation since October 2014. It can acquire, process and analyze high-resolution SAR imagery enabling to detect non-cooperative fishing vessels via a Vessel Detection System (VDS). In these radar images, metallic vessels appears as bright echoes that can be detected via image processing. In addition, the use of satellite Automatic Identification System (satAIS) tracking data completes the solution by providing maritime awareness for most of vessels including larger vessels such as tankers or cargo. It is of particular interest in areas with mixed traffic where the VDS/VMS is not sufficient to identify IUU activities. The integration of such data was supported by a set of R & D programs such as Greidanus (2007) and EU (2013). In this paper, the benefit of SAR-based VDS in completing VMS over the Arafura Seas is presented. This study does not consider unreported and unregulated fishing activities, but solely illegal fishing. In particular, an estimate of the amount of illegal fishing activities via a systematic correlation between VMS-tracked fishing vessels and SAR-based echoes is provided. By integrating Sat-AIS data in the analysis, and thus
removing the SAR echoes corresponding to other type of vessels (e.g. cargo, tankers), non-correlated SAR echoes are potentially illegal fishing vessels. Section 2 describes the newly operational INDESO system. In particular, the interest of Sat-AIS and SAR-based VDS are highlighted. The joint analysis of AIS, SAR-based VDS reports and VMS data is then performed via a systematic matching process. An attempt to provide the share of illegal fishing activities is provided in Section 3. Section 4 concludes this study. 2. Description of INDESO system and subsystems 2.1. VMS data The Indonesian MCS strategy strongly relies on data analysis performed by the FMC where VMS data have been originally handled. It is actually recognized as a cornerstone for any fisheries MCS system. VMS is used by national regulatory authorities, and is not public such that sensitive positional information about fishing areas are not available to other fishing vessels. In Indonesia, it has been activated since 2003 when national authorities noticed a significant decrease in fish stocks and in the mean size of individual species. A description of the VMS system in Indonesia is provided in Suhendar (2013). To date, VMS messages are sampled every 60 min (24 positions per day plus eventually 12 complementary upon specific programming) with information on vessel speed and route, eventually log-book, assistance messages or other information such as estimated arrival time. Several studies have also demonstrated the relevance of VMS trajectories data to monitor fishing activities such as presence of a vessel within exclusion zones, gear types and fishing methods (Bertrand et al., 2007; Bez et al., 2011; Joo et al., 2011; Marzuki et al., 2015). In the following, VMS data made available from October 2014 to March 2015, and in January 2016 are used. 2.2. Rationale The new concept of IUU monitoring is based on the joint use of VMS and VDS based on SAR imagery. All vessels (above a reasonable size) can be detected by images provided by SAR sensors over a monitored area, independently on the cloud cover and light condition (day/night). Cooperative fishing vessels transmit VMS signals while non-cooperative vessels are spotted by radar imagery with no corresponding VMS positions. Non-cooperative vessels are more likely to be poachers. Based on the joint analysis of VMS and VDS data, law enforcement stakeholders can plan a surveillance operation and send a patrol boat to 2
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
Time responsiveness is a key element in maritime surveillance, so that maritime patrols can be associated with satellite radar detections to optimize IUU surveillance operations. For the automatic processes, radar images are produced, indexed by the system, and automatic ship detection and characterization are made available in about 10 min after sensing. The results are then available for further analysis and manual editing by trained image experts in the INDESO operation room. As soon as the VDS report is available (about 10–20 min later), a dedicated GIS platform automatically ingests the ship detection report and performs the automatic correlation with VMS data that have been previously ingested. For the association process, the VMS data are interpolated between position reports at times either side of the image collection. The association process can lead to three possible situations:
inspect the suspected target. This is an effective solution to arrest violating boats without any license or not equipped with VMS. The high cost of satellite radar data is a potential issue for sustained operational. The Agency for Marine and Fisheries Research and Development (AMFRD) of MMAF has recently implemented its own infrastructure operated since October 2014 via the INDESO project. The service provided by the INDESO infrastructure uses the vessel detection reports generated in Near-Real Time (NRT - typically in less than 30 min after each satellite pass) at the newly built satellite receiving station located in Perancak, Bali, Indonesia. Vessel detection reports' are then merged with the VMS data into a dedicated GIS system to allow trained operators to perform fishing detection using both ad hoc matching algorithms and contextual information. Whereas fishing vessels in the European Union have been gradually equipped with AIS transmitters (since May 2014, all vessels over 16 m), no such regulation exists for Indonesian waters. However, as stated by the International Maritime Organization, AIS transmitters shall be fitted aboard all ships of 300 gross tonnage and upwards engaged on international voyages, cargo ships of 500 gross tonnage and upwards not engaged on international voyages and all passenger ships irrespective of size. AIS data may therefore be of high interest to track some foreign fishing vessels, and correlate VDS report with non-fishing vessels. At this stage, it should be noted that Long Range Identification Tracking (LRIT) system is not used our study. This system which apply to certain type of ships engaged on international voyages report the positions of vessels to their flag administration only.
• The detected SAR echo is associated with one VMS vessel, which is the nominal case for fishing vessel; • The detected SAR echo is associated with one AIS vessel, which is •
The automatic vessel detection is based on state-of-the-art algorithms described in Hajduch and Kerbaol (2006) with detailed descriptions of assessment strategy in Longépé et al. (2013) and Pelich et al. (2015). The incidence angle strongly impacts the detection capabilities, the larger the better. However, the number of false alarms may increase as well. From our experience, we estimate that, after an analysis by trained operators, less than 10% of detected echoes remain false alarms for Fine Wide imagery sensed in between 31° to 39°. This ratio increases to about 15% for incidence angle in between 39° to 45°. This remains a challenge for trained operators, especially in the context of NRT services with limited amount of time. Breaking waves, radar-processing ambiguities artifacts which are sometimes tricky to discriminate (especially range ambiguities from land more than hundreds kilometers away - especially true for Indonesian archipelagos) and radar image quality may impact the quality of Vessel Detection Reports. For detailed explanations of these artifacts, the reader may refer to Crisp (2004). To assess the detection rate of fishing vessels with the Wide Fine acquisitions, a previous study (Pelich et al., 2015) is used as a proxy. The detection rate depends on many parameters which are beyond the scope of this study (inter alia sea state & wind speed, incidence angle, the number of looks, the polarization, the pixel spacing/sampling, the radar resolution, the radar cross section of vessel, the frequency of electromagnetic waves...). Here, the assessment is simplified assuming the detection rate depends on the pixel spacing (proxy of radar resolution by a factor 2) and the incidence angle only. From Pelich et al. (2015), the rate of good detection for other sensing modes of the Radarsat-2 satellite is modified considering 1) the new ratio of vessel size w.r.t. pixel spacing and 2) a re-normalization of the impact of the incidence angle via the Figs. 4 and 5 of Pelich et al. (2015) and the ones of our dataset. The estimated rates of good detection with respect to vessel length are provided in Fig. 2.
2.3. VDS with high resolution radar satellites The VDS is based on the capacity to receive and process high resolution satellite radar imagery in NRT via a dedicated satellite receiving station. The satellite receiving antenna (5.4 m diameter) and its processor terminal can receive radar images from two types of satellites, namely the Canadian Radarsat-2 and the Italian CosmoSkyMed satellites. Monitoring requirements are translated into satellite acquisition plans and sent to the satellite operating agency. This task is performed by the INDESO operations on a quarterly basis. In case of surveillance operations at sea, it is of the utmost importance that the route of the patrol vessels is guided by the satellite acquisition plan and not the reverse since there is no way to change the satellite orbits. From October 2014 to February 2017, 896 Radarsat-2 and 130 CosmoSkyMed images were received, processed and analyzed in NRT at INDESO station and 960 ship detection reports were delivered. For Radarsat-2 satellites, two modes (so called Wide and Wide Fine) are mostly acquired as they offer the capabilities to detect small vessels over relatively large area. As of end 2016, the current rate of SAR image acquisitions is about 28 WideFine Radarsat-2 images per month. In this study, SAR data acquired over the Arafura Sea concomitantly with available VMS data are used, i.e. from October 2014 to March 2015, and in January 2016. For the first and second periods of analysis, 139 and 9 images are used, respectively. 120 among these 148 images have been acquired with incidence angle ranging from 31° to 39°. The main characteristics of these two modes are provided in Table 1. Table 1 Characteristics and availability of Radarsat-2 imagery in the INDESO database with available VMS data. The abbreviations rg. and az. stand for the range and azimuth directions, respectively. Beam mode
Wide Wide fine
Nb. images
Coverage
Pixel spacing
Radar resolution
2014–2015/ 2016 11/0 128/9
rg. × az. (km)
rg. × az. (m)
rg. × az. (m)
150 × 150 150 × 170
12.5 × 12.5 6.25 × 6.25
40.0–19.2 × 24.7 15.2–8.2 × 7.7
the nominal case for non-fishing vessel. For instance, the sat-AIS data over the Arafura Sea contain a very limited number of fishing vessels, but mostly tankers or cargoes; The detected SAR echo is associated with neither AIS nor VMS vessels. Depending on whether its length estimated from the SAR image is below 50 m, it may be considered as a suspected fishing vessel.
2.4. Sat-AIS The AIS uses the maritime Very High Frequency (VHF) band for the transmission and reception of data. It was first implemented as a technology to avoid collisions among large vessels at sea and exchange information such as identification, position, course and speed. AIS messages are sent from all merchant ships above 300 gross tons engaged in international voyages, ships above 500 GT in domestic 3
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
terrestrial AIS receivers are installed onshore. For class-B vessel, this technology may be less effective as the transmit power is rather low and close to the noise floor of the system. A comprehensive overview of AIS technology is provided in Carson Jackson (2012). As opposed to VMS, AIS data are aimed at being publicly available as a commercial product. The introduction of error, the falsification of the AIS messages or spoofing attacks are increasingly found:
• Non-intentional errors can be made by manually entering data. It •
Fig. 2. Estimated good detection rate of vessels from Radarsat-2 fine wide mode, depending upon the range of incidence angles of the acquired image.
•
voyages, and all passenger vessels. Most of large vessels are equipped with a so-called Class-A transceiver (transmit and receive) with 12.5 W transmit power and the capability to reserve transmit slot via 2 receivers in continuous operation. Class-A AIS transponder broadcasts positional messages every 2 to 10 s while underway, and every 3 min while at anchor. For leisure vessel and other smaller vessel, lighter solution - so called class-B - exists with less transmit power (2 W) and less sophisticated solution for transmit slot identification. The broadcast rate is nominally every 30 s or 3 min depending on vessel speed. Beyond position coordinates, AIS data contains Maritime Mobile Service Identities (MMSI) information including Maritime Identification Digits which are a direct identifier for vessel's flag. More specifically, the first digit indicates the region of origin with “3”, “4”, and “6”, the “North and Central America and Caribbean”, “Asia”, and “Africa”, respectively. While AIS is originally a “line of sight” technology with a coverage limit of about 20 nautical miles around the spherical Earth with vesselto-vessel or vessel-to-terrestrial transceivers, satellite-based AIS receivers is now widely used to ensure data reception even when no
may concern destination, vessel type but also MMSI information, 2% of cases according to Iphar et al. (2015). The falsification in the messages is clearly an intentional action aiming at hindering any potential surveillance systems. The most common falsification is to turn off AIS transmitters, which may hide illicit activities. It is important to note that turning off AIS is not, in and of itself, evidence of illicit activity. In some specific areas, this may be used to protect from pirates or to protect commercially sensitive information such as fishing grounds. The spoofing of messages is a voluntary action “from an actor which is external to the vessel, in order to mislead both the crew on board and the outside world on the behavior of a vessel.” (Iphar et al., 2015).
The first two items severely affect the potential of AIS to provide a comprehensive maritime awareness, not to mention the time latency and time revisit. New AIS-receiver satellite constellation has recently started to unroll from beginning of 2017 with claimed latencies and revisit of few minutes. As an illustration of the above behaviors, sat-AIS data from the Arafura Sea are analyzed. More specifically, the period before and after the establishment of the moratorium by MMAF's Regulation 56/2014 is studied. Since November 2014, foreign fishing vessels over 30 GT must renew their fishing license annually, and intensive investigation on suspected ex- foreign vessel has been conducted. Whereas the number of AIS messages transmitted by all the other types of vessel fluctuate with no real trend from November 2014, the number of vessels identified as “Fishing Vessel” drastically decreased from October to December 2014 (see blue curve in Fig. 3b). Fig. 3. The averaged number of vessels transmitting on AIS, and their declared type in the Arafura Sea: large vessels (a) and smaller vessels (b). The dash vertical line indicates the establishment of the MMAF moratorium for ex-foreign fishing vessel. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
4
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
“Leisure ship” while the pictures associated to their MMSI indicate fishing vessel. As given by VMS license, most of the fishing vessels are smaller than 30 m in the area of interest, few of them being however larger up to 50 m. In the area of interest, few other types of vessels are smaller than 50 m (see Fig. 3b). To decipher fishing from non-fishing vessels from SAR imagery, the estimated length of the detected echoes can thus be used as a decision rule, based on the good reliability of this estimator. To this end, the sizes of SAR-based echoes are compared with actual vessel length from VMS fishing vessels via a VMS-SAR matchups database. The results are shown in Fig. 7. Only 11 vessels among 515 have an estimated length over 50 m (which represent only 2.1% of the cases). This amount is considered as negligible in the following calculation. As a consequence, all SAR-based vessels whose estimated length is below 50 m are assumed to be fishing vessels. Considering all the above assumptions, the ratio of illegal fishing Rillegal can be estimated based on the assumption that legal fishing vessels carry VMS transceiver (see Fig. 6). With Nillegal being the number of illegal fishing vessels among all Nfish fishing vessels, we have from October 2014 to April 2015:
Fig. 4. Suspected vessels in the Arafura Sea via an analysis of MMSI. The dash vertical line indicates the establishment of the MMAF moratorium for ex-foreign fishing vessel.
Via an analysis of fishing vessels' MMSI for all AIS messages received from July 2014 to April 2016, most of fishing vessels had Asian flags (31 over 38). Only two had MMSI indicating North and Central America and Caribbean flag with over 1500 GT and 70 m length. One was registered from Western Africa. Last but not least, four of them had “fake” MMSI not following the rules of the International Telecommunication Union (e.g. 100000000, 110000000, 123000000, 123456789). As illustrated in Fig. 4, all vessels with MMSI starting with “3”, “4” or “6” actually left the Arafura Sea due to the moratorium. At this stage, two assumptions can be raised 1) these vessels went fishing in other seas or 2) they adapted their AIS transceiver filling erroneous MMSI or vessel type information. In any cases, the moratorium has clearly induced a change of behavior for foreign fishing fleets. The Sat-AIS data cannot provide on its own the solution for an efficient monitoring system for fishing vessels. In the following, it is essentially used to identify large vessels such as cargo or tankers.
⎧ Nillegal ⎨ Nfish ⎩
= 2 + 2 + 1 + (671 − 67) = 609 = 736 + (671 − 67) + 2 + 2 + 1 = 1345
A a result, approximately 45.3% of fishing vessels in the area are illegal (in the sense that they are not compliant with VMS) during the first period of analysis (from Oct. to April 2015). In January 2016 with the few data available, this amount is estimated to be 36.1%. 3.2. Refined estimates The rough estimate does not take into account bias and errors due to the proposed methodologies. The potential issues are numerous, linked to the SAR-based detection, to the temporal sampling of AIS and/or VMS, the methodologies to interpolate AIS/VMS tracks and then to correlate with SAR-based echoes. In the following, some issues are considered, discussed, and their impact on the above estimate evaluated. As indicated by Fig. 2, the probability of SAR-based detection pd for vessels with length above 15 m but below 50 m ranges from 80 to 100%. Considering that the fishing vessels that were not detected by the SAR detector are partly legal and illegal with exactly the same ratio as of all the other fishing vessels, we thus have
3. Towards the assessment of illegal fishing In this section, an assessment of the proportion of illegal fishing activities is provided. To do so, the cross-analysis of SAR-derived vessel detections and VMS/AIS tracks is used from the 148 Radarsat-2 images mentioned in Section 2.3. AIS and VMS tracks are interpolated at the time of each SAR acquisition time, and interpolated echoes are then matched with corresponding SAR echoes following distance criteria (see Figs. 5 and 6). A post-processing analysis is carefully operated via GIS software to remove any erroneous data, e.g. close to shoreline or corresponding to mooring areas.
⎧ Nillegal = 609 + pillegal (NfishSAR/ pd − NfishSAR)) ⎨ Nfish = 1345 + (NfishSAR/ pd − NfishSAR) ⎩ with NfishSAR = (671 − 67) + 538 + 10 + 4 = 1156 being the number of fishing vessels as detected by the SAR detector. After solving this equation, the probability of detection pd does not affect the computation of illegal fishing rate. In addition, the false alarm rate for the SAR-based vessel detector should be considered as mentioned in Section 2.3. Here, a false alarm rate RFA ranging from 5 to 15% for all vessels below 50 m is assumed. Above 50 m, no false alarm is considered. It is assumed that false alarms only impact SAR targets non-correlated with VMS echoes, SAR targets associated to VMS being “true” vessels. A set of falsely (671 − 67)RFA echoes should then be removed from the statistics, the ratio of illegal fishing vessel becomes
3.1. Evaluation principles Before any attempt to assess the share of illegal over legal fishing from this data, a set of assumptions is considered. The VMS equipped vessels are fishing vessels and all of them are fishing legally. This hypothesis is an overestimation of the fleet of legal fishing vessels as vessel can transmit their position using VMS technology while having an invalid or expired fishing license. In addition, being VMS compliant does not preclude from having legal fishing activities. At this stage, it should be recall that this study does not consider unreported and unregulated fishing activities. With respect to AIS, the vessels reporting their AIS types as “fishing” are actually fishing vessels. The “type” field of AIS message does not necessarily reflect the real type of the vessel. However, the vessels fishing legally do not have reason to spoof this field whereas the vessels fishing illegally may have reason to spoof it. This can be refined by manual inspection of the database of each unique vessel. For instance, two potential fishing vessels are spotted as they report their type as
Rillegal =
609 − 604RFA 1345 − 604RFA
Vessels equipped with VMS transmit their positions regularly every 60 min. While performing the matching process, the time lag between one of the two VMS messages and the SAR echoes should be thus 30 min maximum. Erroneous VMS data with sampling exceeding 60 min are discarded. In addition, the maximum distance between SAR 5
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
Fig. 5. Map of SAR-VMS-AIS matchups from INDESO database (from October 2014 to April 2015, and in January 2016). Red square: VMS unmatched, green square: VMS-SAR matchups, blue circle: AIS unmatched, pink circle: AIS-SAR matchups, and black star: SAR unmatched. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
and interpolated VMS positions is heightened as 20 km to ensure the highest number of matchups. However, a fraction of them might be considered as erroneous, thus adding a certain number of SAR-based targets as potential illegal fishing vessel. This mismatch rate Rmis may range up to 20% out of the 538 matchups. From October 2014 to April 2015, the ratio of illegal fishing vessel thus depends on the SAR false alarms rate and the rate of mismatch between VMS and SAR:
Rillegal =
609 − 604RFA + 538Rmis Nfish/ Nvessel 1345 − 604RFA + 538Rmis Nfish/ Nvessel
where Nvessel is the number of all vessels in the areas observed by the SAR imageries (incl. tankers...). It is equal to Nvessel = 736 + 157 + 46 + 671 = 1610. The above equation and the one corresponding to January 2016 are illustrated by Fig. 8. From October 2014 to April 2015, the estimated share of illegal fishing activities with respect to the total fishing activities in the Arafura Sea ranges from about 42 to 47% with RFA in the [5–15]% interval. In January 2016, it ranges from about 32 to 42%. At this stage, it is recall
Fig. 7. Assessment of vessel length estimation based on 515 SAR-VMS matchups and license information from MMAF - among the 738 matchups, some vessel lengths were missing in the VMS MMAF license information.
Fig. 6. Overview of SAR-VMS-AIS matchups from INDESO database: from October 2014 to April 2015, corresponding to 139 analyzed SAR images (left), and in January 2016 corresponding to 9 analyzed SAR images (right).
6
Marine Pollution Bulletin xxx (xxxx) xxx–xxx
N. Longépé et al.
Acknowledgments This study has been funded by the INDESO project and the Indonesian Ministry of Fisheries and Marine Affairs. In addition, some technical processes (sat-AIS data manipulation, SAR/VDS coupling) were partly co-funded by ESA/ESTEC via the AMTRAC project and ARTES 20 program. We would like to thank the Editor and the two anonymous reviewers for their fruitful comments and suggestions. References Agnew, J., Pearce, J., Ganapathiraju, P., Peatman, T., Watson, R., Beddington, J., Pitcher, T., 2009, 02. Estimating the worldwide extent of illegal fishing. PLoS One 4, 1–8. Bertrand, S., Bertrand, A., Carrasco, R.G., Gerlotto, F., 2007. Scale invariant movements of fishermen: the same foraging strategy as natural predators. Ecol. Appl. 17, 331–337. Bez, N., Walker, E., Gaertner, D., Rivoirard, J., Gaspar, P., 2011. Fishing activity of tuna purse seiners estimated from vessel monitoring system (VMS) data. Can. J. Fish. Aquat. Sci. 68, 1998–2010. Carson Jackson, J., 2012. Satellite AIS - developing technology or existing capability? J. Navig. 6, 303–321. Claus, S., De Hauwere, N., Vanhoorne, B., Dias, S., Oset García, P., Hernandez, F., Mees, J., 2017. Database marine regions. Technology Fact Sheets. Crisp, D., 2004. The state-of-the-art in ship detection in Synthetic Aperture Radar imagery. In: DSTO. EU, 2013. Final report summary - DOLPHIN (development of pre-operational services for highly innovative maritime surveillance capabilities). In: Technical Report. Project ID: 263079. FAO, 1995. Code of conduct for responsible fisheries (CCRF). In: Technical Report. Food and Agriculture Organisation of the United Nations, ROME. FAO, 2012. The state of world fisheries and aquaculture. In: Technical Report. FAO Fisheries and Aquaculture Department. FAO, 2014. The State of World Fisheries and Aquaculture: Opportunities and Challenges. 978-92-5-108275-1 (E-ISBN 978-92-5-108276-8, Rome. Greidanus, H., 2007. DECLIMS: detection, classification and identification of marine traffic from space: final report. In: Technical Report. EC-JRC. Hajduch, G., Kerbaol, V., 2006. Ships detection on envisat asar data: results, limitations and perspectives. In: SEASAR 2006, ESA/ESRIN, Frascati. Husein, S.Y., 2015. Strategy on combating IUU fishing and post moratorium policies plan. In: FishCRIME, Cape Town, South Africa. Iphar, C., Napoli, A., Ray, C., 2015, October, October. Detection of false AIS messages for the improvement of maritime situational awareness. In: Oceans2015, Washington, DC, United States. Joo, R., Bertrand, S., Chaigneau, A., iquen, M., 2011. Optimization of an artificial neural network for identification of fishing event positions from vessel monitoring system data. Ecol. Model. 222, 1048–1059. Kourti, N., Shepherd, I., Schwartz, G., Pavlakis, P., 2001. Integrating spaceborne SAR imagery into operational systems for fisheries monitoring. Can. J. Remote. Sens. 4, 291–305. Lehodey, P., Senina, I., Wibawa, T., Titaud, O., Calmettes, B., Conchon, A., Tranchant, B., Gaspar, P., 2017. Operational modelling of bigeye tuna (Thunnus obesus) spatial dynamics in the Indonesian region. Mar. Pollut. Bull (in press). Longépé, N., Hajduch, G., Pelich, R., Habonneau, J., Lebras, J., 2013. SAR-based ship monitoring: advanced methodologies with medium resolution images (from WSM ASAR to EWS/IWS S1 mission). In: ESA Living Planet Symposium. Marzuki, M.I., Gaspar, P., Garello, R., Kerbaol, V., Fablet, R., 2015. Fishing gear recognition from VMS data to identify illegal fishing activities in Indonesia. In: OCEANS 2015 - Genova. Nurhakim, S., Nikijuluw, V., Badrudin, M., Pitcher, T., Wagey, G., 2008. A study of illegal, unreported and unregulated (IUU) fishing in the Arafura Sea, Indonesia. In: Technical Report. FAO, Rome. Pelich, R., Longépé, N., Mercier, G., Hajduch, G., Garello, R., 2015. AIS-based evaluation of target detectors and SAR sensors characteristics for maritime surveillance. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 8, 3892–3901. Petrossian, G., 2015. Preventing illegal, unreported and unregulated (IUU) fishing: a situational approach. Biol. Conserv. 189, 3948. Purwanto, 2013, September. Fishing fleet productivity and potential production of shrimp fishery in the Arafura Sea. J. Lit. Perikan. Ind. 19, 147–155. Resosudarmo, B., Napitupulu, L., Campbell, D., 2009. Working With Nature Against Poverty: Development, Resources and the Environment in Eastern Indonesia: Illegal Fishing in the Arafura Sea. ISEASYusof Ishak Institute. Sodik, D.M., 2007. Combating illegal, unreported and unregulated fishing in Indonesian waters: the need for fisheries legislative reform. University of Wollongong Ph.D. thesis. Suhendar, M., 2013. Comparison of vessel monitoring system (VMS) between iceland and indonesia [final project]. In: Technical Report. Ministry of Marine Affairs and Fisheries.
Fig. 8. Estimated rate of illegal fishing activities in the Arafura Sea based on VMS-SARAIS matchups for the two analyzed periods: from October 2014 to April 2015 with solid line, and January 2016 with dashed line.
this estimate is based on the major assumption that all VMS equipped vessels are legal with no consideration on their associated license and their activity. The joint analysis of the VMS/SAR/AIS correlation and VMS-related license data is left to another study. Some vessels may breach the conditions of their licenses (Sodik, 2007). From data extracted in 2011, Purwanto (2013) stated the amount of unlicensed shrimp trawl vessels is about 63%. However, the number of shrimp trawlers compared to fish trawlers is low (in 2005, 301 compared to 842 (Purwanto, 2013)), not to mention bottom long lines vessels. The estimated amount of illegal fishing activities from this study should not be compared to this amount, the period of analysis being different as well. With a new moratorium decided by the Indonesian government from November 2014, and the operational phase of the INDESO system starting from October 2014, illegal fishing activities should be decreasing, as it is observed from this study. 4. Conclusion In this study, the benefit of a complete MCS system is demonstrated. Especially, the added-value of AIS and SAR-based VDS complementary to that provided by VMS is exposed. The INDESO project provides the Indonesian MCS with the capacity to acquire and process SAR imagery. With its NRT capability, SAR-based VDS combined with VMS data and patrol vessels is a solution to prevent, deter and eliminate IUU fishing. In our analysis, the joint analysis of VMS-AIS and VDS report has been carried out, leading to an estimated amount of illegal fishing in the Arafura Sea based on the assumption that legal fishing vessels carry VMS systems. The considered period is mainly from October 2014 to March 2015, i.e just after the establishment of new moratorium by the MMAF in November 2014. It has been estimated that non-cooperative fishing vessels represent between 42 and 47% of the total number of fishing vessels in the Arafura Sea at that period of time. Even though the number of data available in January 2016 is limited, the share of illegal fishing activities seems lower almost a year later (up to −15% reaching potentially 32%). As stated in Husein (2015), a task force on Prevention and Eradication of IUU established by the MMAF leads to the compliance audit of 1132 ex-foreign vessels from March to October 2015. This audit plus a series of law enforcement actions supported by the INDESO infrastructure and relayed by mass media during 2015 surely impacted the amount of illegal but also unreported and unregulated fishing. The expected decrease should be further consolidated and monitored continuously afterwards. This research provides a baseline for future monitoring.
7