Offshore fixed infrastructure is a global dataset that uses AI and machine learning to detect and classify structures throughout the world’s oceans.
Classification labels (oil, wind, and unknown) are provided, as well as confidence levels (high, medium, or low) reflecting our certainty in the assigned label. Detections can be filtered and colored on the map using both label and confidence level.The data is updated on a monthly basis, and new classified detections are added at the beginning of every month. Viewing change using the timebar is simple, and allows anyone to recognize the rapid industrialization of the world’s oceans. For example, you can easily observe the expansion of wind farms in the North and East China Seas, or changes in oil infrastructure in the Gulf of Mexico or Persian Gulf.
By overlaying the existing map layers, you can explore how vessels interact with oil and wind structures, visualise the density of synthetic aperture radar (SAR) and Visible Infrared Imaging Radiometer Suite (VIIRS) vessel detections around infrastructure, or determine which marine protected areas (MPAs) contain wind, oil, or other infrastructure types. These are only examples of the types of questions we can now ask. Offshore fixed infrastructure is a first of its kind dataset that not only brings to light the extensive industrialization of our oceans, but enables users across industries to use this information in research, monitoring and management.
We use SAR imagery from the Copernicus Sentinel-1 mission of the European Space Agency (ESA) [1]. The images are sourced from two satellites (S1A and S1B up until December 2021 when S1B stopped operating, and S1A only from 2022 onward) that orbit 180 degrees out of phase with each other in a polar, sun-synchronous orbit. Each satellite has a repeat-cycle of 12 days, so that together they provide a global mapping of coastal waters around the world approximately every six days for the period that both were operating. The number of images per location, however, varies greatly depending on mission priorities, latitude, and degree of overlap between adjacent satellite passes. Spatial coverage also varies over time [2]. Our data consist of dual-polarization images (VH and VV) from the Interferometric Wide (IW) swath mode, with a resolution of about 20 m.
[1] https://sedas.satapps.org/wp-content/uploads/2015/07/Sentinel-1_User_Handbook.pdf
[2] https://sentinels.copernicus.eu/web/sentinel/missions/sentinel-1/observation-scenario
Detecting infrastructure with SAR is based on the widely used Constant False Alarm Rate (CFAR) algorithm, an anomaly detection method conceived for detecting ships in synthetic aperture radar images, that has been modified to remove non-stationary objects. This algorithm is designed to search for pixel values that are unusually bright (the targets) compared to those in the surrounding area (the sea clutter). This method sets a threshold based on the pixel values of the local background (within a window), scanning the whole image pixel-by-pixel. Pixel values above the threshold constitute an anomaly and are likely to be samples from a target.
To classify every detected offshore infrastructure, we used deep learning and designed a ConvNet based on the ConvNeXt architecture. A novel aspect of our deep learning classification approach is the combination of SAR imagery from Sentinel-1 with optical imagery from Sentinel-2. From six-month composites of dual-band SAR (VH and VV) and four-band optical (RGB and NIR) images, we extracted small tiles for every detected fixed infrastructure, with the respective objects at the center of the tile. A single model output includes the probabilities for the specified classes: wind, oil, unknown, lake maracaibo, and noise.
GFW post-processed the classified SAR detections to reduce noise (false positives), remove vessels, exclude areas with sea ice at high latitudes, and incorporate expert feedback. We used a clustering approach to identify detections across time (within a 50 m radius) that were likely the same structure but their coordinates differed slightly, and assigned them the greatest average predicted label of the cluster. We also filled in gaps for fixed structures that were missing in one timestep but detected in the previous and following timesteps, and dropped detections appearing in a single timestep. Finally, the dataset underwent extensive manual review and editing by researchers and industry experts in order to refine the final product, and provide the most accurate dataset possible.
Each detection has a unique individual identifier (detection_id). A six-month image composite is used in the classification, therefore the detection_date represents the middle of the six month period. This helps to remove non-stationary objects (i.e. vessels), and avoid confusion in the model if a structure is being built, or there isn’t adequate imagery available. structure_id allows you to track a structure through time. There are therefore many detection_id (one for each month the structure is detected) for each structure_id. Labels of wind and oil represent any wind or oil related structure respectively. Unknown represents a structure that is not oil or wind related, such as bridges or navigational buoys.
Label confidence levels of ‘High’. ‘Medium’ and ‘Low’ are assigned to each structure, and are conditional on where the detections fell in relation to the boundaries of manually developed wind and oil polygons, and whether the label has changed from the previous month. The label_confidence field can be used to filter analysis.
Two repos are used in the automation process, both of which are private, and should not be shared publicly.
Detection and classification: https://github.com/GlobalFishingWatch/sentinel-1-ee/tree/master
Clustering and reclassification: https://github.com/GlobalFishingWatch/infrastructure-post-processing
All code developed for the paper, Paolo, F.S., Kroodsma, D., Raynor, J. et al. Satellite mapping reveals extensive industrial activity at sea. Nature 625, 85–91 (2024). https://doi.org/10.1038/s41586-023-06825-8, including SAR detection, deep learning models, and analyses is open source and freely available at https://github.com/GlobalFishingWatch/paper-industrial-activity.
Copernicus Sentinel data 2017-current
Lujala, Päivi; Jan Ketil Rød & Nadia Thieme, 2007. 'Fighting over Oil: Introducing A New Dataset', Conflict Management and Peace Science 24(3), 239-256
Sabbatino, M., Romeo, L., Baker, V., Bauer, J., Barkhurst, A., Bean, A., DiGiulio, J., Jones, K., Jones, T.J., Justman, D., Miller III, R., Rose, K., and Tong., A., Global Oil & Gas Infrastructure Features Database Geocube Collection, 2019-03-25, https://edx.netl.doe.gov/dataset/global-oil-gas-infrastructure-features-database-geocube-collection, DOI: 10.18141/1502839
Non-Commercial Use Only. The Site and the Services are provided for Non-Commercial use only in accordance with the CC BY-NC 4.0 license. If you would like to use the Site and/or the Services for commercial purposes, please contact us.
Infrastructure development methods should reference the paper:
Paolo, F.S., Kroodsma, D., Raynor, J. et al. Satellite mapping reveals extensive industrial activity at sea. Nature 625, 85–91 (2024). https://doi.org/10.1038/s41586-023-06825-8
All code developed for the paper, including SAR detection, deep learning models, and analyses is open source and freely available at https://github.com/GlobalFishingWatch/paper-industrial-activity. All the data generated and used by these scripts can reference the following data repos:
Analysis and Figures: https://doi.org/10.6084/m9.figshare.24309475
Training and Evaluation: https://doi.org/10.6084/m9.figshare.24309469
", "schema": { "label": { "keyword": "label", "enum": { "oil": "oil", "wind": "wind", "unknown": "unknown" } }, "structure_id": "structure_id", "label_confidence": { "keyword": "label_confidence", "enum": { "high": "high", "low": "low", "medium": "medium" } }, "structure_end_date": "structure_end_date", "structure_start_date": "structure_start_date" } }, "public-fixed-infrastructure-filtered": { "name": "Offshore Fixed Infrastructure (SAR, Optical)", "description": "Offshore fixed infrastructure is a global dataset that uses AI and machine learning to detect and classify structures throughout the world’s oceans.
Classification labels (oil, wind, and unknown) are provided, as well as confidence levels (high, medium, or low) reflecting our certainty in the assigned label. Detections can be filtered and colored on the map using both label and confidence level.The data is updated on a monthly basis, and new classified detections are added at the beginning of every month. Viewing change using the timebar is simple, and allows anyone to recognize the rapid industrialization of the world’s oceans. For example, you can easily observe the expansion of wind farms in the North and East China Seas, or changes in oil infrastructure in the Gulf of Mexico or Persian Gulf.
By overlaying the existing map layers, you can explore how vessels interact with oil and wind structures, visualise the density of synthetic aperture radar (SAR) and Visible Infrared Imaging Radiometer Suite (VIIRS) vessel detections around infrastructure, or determine which marine protected areas (MPAs) contain wind, oil, or other infrastructure types. These are only examples of the types of questions we can now ask. Offshore fixed infrastructure is a first of its kind dataset that not only brings to light the extensive industrialization of our oceans, but enables users across industries to use this information in research, monitoring and management.
We use SAR imagery from the Copernicus Sentinel-1 mission of the European Space Agency (ESA) [1]. The images are sourced from two satellites (S1A and S1B up until December 2021 when S1B stopped operating, and S1A only from 2022 onward) that orbit 180 degrees out of phase with each other in a polar, sun-synchronous orbit. Each satellite has a repeat-cycle of 12 days, so that together they provide a global mapping of coastal waters around the world approximately every six days for the period that both were operating. The number of images per location, however, varies greatly depending on mission priorities, latitude, and degree of overlap between adjacent satellite passes. Spatial coverage also varies over time [2]. Our data consist of dual-polarization images (VH and VV) from the Interferometric Wide (IW) swath mode, with a resolution of about 20 m.
[1] https://sedas.satapps.org/wp-content/uploads/2015/07/Sentinel-1_User_Handbook.pdf
[2] https://sentinels.copernicus.eu/web/sentinel/missions/sentinel-1/observation-scenario
Detecting infrastructure with SAR is based on the widely used Constant False Alarm Rate (CFAR) algorithm, an anomaly detection method conceived for detecting ships in synthetic aperture radar images, that has been modified to remove non-stationary objects. This algorithm is designed to search for pixel values that are unusually bright (the targets) compared to those in the surrounding area (the sea clutter). This method sets a threshold based on the pixel values of the local background (within a window), scanning the whole image pixel-by-pixel. Pixel values above the threshold constitute an anomaly and are likely to be samples from a target.
To classify every detected offshore infrastructure, we used deep learning and designed a ConvNet based on the ConvNeXt architecture. A novel aspect of our deep learning classification approach is the combination of SAR imagery from Sentinel-1 with optical imagery from Sentinel-2. From six-month composites of dual-band SAR (VH and VV) and four-band optical (RGB and NIR) images, we extracted small tiles for every detected fixed infrastructure, with the respective objects at the center of the tile. A single model output includes the probabilities for the specified classes: wind, oil, unknown, lake maracaibo, and noise.
GFW post-processed the classified SAR detections to reduce noise (false positives), remove vessels, exclude areas with sea ice at high latitudes, and incorporate expert feedback. We used a clustering approach to identify detections across time (within a 50 m radius) that were likely the same structure but their coordinates differed slightly, and assigned them the greatest average predicted label of the cluster. We also filled in gaps for fixed structures that were missing in one timestep but detected in the previous and following timesteps, and dropped detections appearing in a single timestep. Finally, the dataset underwent extensive manual review and editing by researchers and industry experts in order to refine the final product, and provide the most accurate dataset possible.
Each detection has a unique individual identifier (detection_id). A six-month image composite is used in the classification, therefore the detection_date represents the middle of the six month period. This helps to remove non-stationary objects (i.e. vessels), and avoid confusion in the model if a structure is being built, or there isn’t adequate imagery available. structure_id allows you to track a structure through time. There are therefore many detection_id (one for each month the structure is detected) for each structure_id. Labels of wind and oil represent any wind or oil related structure respectively. Unknown represents a structure that is not oil or wind related, such as bridges or navigational buoys.
Label confidence levels of ‘High’. ‘Medium’ and ‘Low’ are assigned to each structure, and are conditional on where the detections fell in relation to the boundaries of manually developed wind and oil polygons, and whether the label has changed from the previous month. The label_confidence field can be used to filter analysis.
Two repos are used in the automation process, both of which are private, and should not be shared publicly.
Detection and classification: https://github.com/GlobalFishingWatch/sentinel-1-ee/tree/master
Clustering and reclassification: https://github.com/GlobalFishingWatch/infrastructure-post-processing
All code developed for the paper, Paolo, F.S., Kroodsma, D., Raynor, J. et al. Satellite mapping reveals extensive industrial activity at sea. Nature 625, 85–91 (2024). https://doi.org/10.1038/s41586-023-06825-8, including SAR detection, deep learning models, and analyses is open source and freely available at https://github.com/GlobalFishingWatch/paper-industrial-activity.
Copernicus Sentinel data 2017-current
Lujala, Päivi; Jan Ketil Rød & Nadia Thieme, 2007. 'Fighting over Oil: Introducing A New Dataset', Conflict Management and Peace Science 24(3), 239-256
Sabbatino, M., Romeo, L., Baker, V., Bauer, J., Barkhurst, A., Bean, A., DiGiulio, J., Jones, K., Jones, T.J., Justman, D., Miller III, R., Rose, K., and Tong., A., Global Oil & Gas Infrastructure Features Database Geocube Collection, 2019-03-25, https://edx.netl.doe.gov/dataset/global-oil-gas-infrastructure-features-database-geocube-collection, DOI: 10.18141/1502839
Non-Commercial Use Only. The Site and the Services are provided for Non-Commercial use only in accordance with the CC BY-NC 4.0 license. If you would like to use the Site and/or the Services for commercial purposes, please contact us.
Infrastructure development methods should reference the paper:
Paolo, F.S., Kroodsma, D., Raynor, J. et al. Satellite mapping reveals extensive industrial activity at sea. Nature 625, 85–91 (2024). https://doi.org/10.1038/s41586-023-06825-8
All code developed for the paper, including SAR detection, deep learning models, and analyses is open source and freely available at https://github.com/GlobalFishingWatch/paper-industrial-activity. All the data generated and used by these scripts can reference the following data repos:
Analysis and Figures: https://doi.org/10.6084/m9.figshare.24309475
Training and Evaluation: https://doi.org/10.6084/m9.figshare.24309469
", "schema": { "label": { "keyword": "label", "enum": { "oil": "oil", "wind": "wind", "unknown": "unknown" } }, "structure_id": "structure_id", "label_confidence": { "keyword": "label_confidence", "enum": { "high": "high", "low": "low", "medium": "medium" } }, "structure_end_date": "structure_end_date", "structure_start_date": "structure_start_date" } }, "public-gfcm-fao": { "name": "GFCM FAO", "description": "A fisheries restricted area (FRA) is a geographically defined area in which some specific fishing activities are temporarily or permanently banned or restricted in order to improve the exploitation patterns and conservation of specific stocks as well as of habitats and deep-sea ecosystems. In the Mediterranean and the Black Sea, 1,760,000 square kilometers of sea habitats are protected by ten FRAs established by the GFCM. This includes one large deep-water FRA (1,730,000 square kilometers) in which the use of towed dredges and trawl nets in all waters deeper than 1000 metres is banned to protect deep-sea benthic habitats. The layer was taken from the GFCM website but manually adjusted to reflect the original coordinates of the FRAs as stipulated in the document REC. GFCM/29/2005/1." }, "public-global-all-tracks": { "name": "Lintasan", "description": "The dataset contains the tracks from all vessels (AIS) - Version 3.0", "schema": { "speed": { "keyword": "speed", "enum": { "0": "0", "20": "20" } } } }, "public-global-all-vessels": { "name": "AIS (All Vessels)", "description": "Vessels from AIS", "schema": { "source": "source" } }, "public-global-bathymetry": { "name": "Batimetri", "description": "Global Fishing Watch classifies an event as an encounter when two vessels are detected:
Users can filter encounter events by vessel characteristics and context, including:
Currently displayed encounter types include:
To view more details about a specific encounter—such as its location or the identity of the encountered vessel—click the “See more” icon associated with the event.
You can read more about transshipment behaviour from our report or scientific publication.
", "schema": { "duration": { "keyword": "duration", "enum": { "2": "2", "48": "48" } }, "next_port_id": "next_port_id", "encounter_type": { "keyword": "encounter_type", "enum": { "CARRIER-FISHING": "CARRIER-FISHING", "SUPPORT-FISHING": "SUPPORT-FISHING", "FISHING-CARRIER": "FISHING-CARRIER", "FISHING-BUNKER": "FISHING-BUNKER", "FISHING-FISHING": "FISHING-FISHING", "CARRIER-BUNKER": "CARRIER-BUNKER", "FISHING-SUPPORT": "FISHING-SUPPORT" } } } }, "public-global-fishing-effort": { "name": "AIS", "description": "Global Fishing Watch uses data about a vessel’s identity, type, location, speed, direction and more that is broadcast using the Automatic Identification System (AIS) and collected via satellites and terrestrial receivers. AIS was developed for safety/collision-avoidance. Global Fishing Watch analyzes AIS data collected from vessels that our research has identified as known or possible commercial fishing vessels, and applies a fishing presence algorithm to determine “apparent fishing activity” based on changes in vessel speed and direction. The algorithm classifies each AIS broadcast data point for these vessels as either apparently fishing or not fishing and shows the former on the Global Fishing Watch fishing activity heat map. AIS data as broadcast may vary in completeness, accuracy and quality. Also, data collection by satellite or terrestrial receivers may introduce errors through missing or inaccurate data. Global Fishing Watch’s fishing presence algorithm is a best effort mathematically to identify “apparent fishing activity.” As a result, it is possible that some fishing activity is not identified as such by Global Fishing Watch; conversely, Global Fishing Watch may show apparent fishing activity where fishing is not actually taking place. For these reasons, Global Fishing Watch qualifies designations of vessel fishing activity, including synonyms of the term “fishing activity,” such as “fishing” or “fishing effort,” as “apparent,” rather than certain. Any/all Global Fishing Watch information about “apparent fishing activity” should be considered an estimate and must be relied upon solely at your own risk. Global Fishing Watch is taking steps to make sure fishing activity designations are as accurate as possible. Global Fishing Watch fishing presence algorithms are developed and tested using actual fishing event data collected by observers, combined with expert analysis of vessel movement data resulting in the manual classification of thousands of known fishing events. Global Fishing Watch also collaborates extensively with academic researchers through our research program to share fishing activity classification data and automated classification techniques.", "schema": { "speed": "speed", "bearing": "bearing", "distance_from_port_km": { "keyword": "Distance from port", "enum": { "0": "0", "1": "1", "2": "2", "3": "3", "4": "4", "5": "5" } } } }, "public-global-fishing-events": { "name": "Fishing Events (AIS)", "description": "The dataset contains fishing events for AIS", "schema": { "type": { "keyword": "type", "enum": { "CARGO": "CARGO", "DISCREPANCY": "DISCREPANCY", "CARRIER": "CARRIER", "FISHING": "FISHING", "GEAR": "GEAR", "OTHER": "OTHER", "PASSENGER": "PASSENGER", "SEISMIC_VESSEL": "SEISMIC_VESSEL", "SUPPORT": "SUPPORT" } } } }, "public-global-fishing-vessels": { "name": "AIS (Fishing Vessels)", "description": "Vessels from AIS", "schema": { "source": "source" } }, "public-global-gaps-events": { "name": "Disabling Events (AIS)", "description": "The dataset contains gaps events for AIS" }, "public-global-loitering-events-carriers": { "name": "Loitering Events for Carriers Vessels (AIS)", "description": "The dataset contains loitering events for AIS (Carriers)." }, "public-global-loitering-events": { "name": "Loitering Events (AIS)", "description": "Global Fishing Watch classifies an event as a loitering event when:
Learn more about considerations of using AIS data by looking at the AIS limitations section in the Apparent fishing events (AIS) data documentation..
", "schema": { "type": { "keyword": "type", "enum": { "CARGO": "CARGO", "DISCREPANCY": "DISCREPANCY", "CARRIER": "CARRIER", "FISHING": "FISHING", "GEAR": "GEAR", "OTHER": "OTHER", "PASSENGER": "PASSENGER", "SEISMIC_VESSEL": "SEISMIC_VESSEL", "SUPPORT": "SUPPORT" } }, "duration": { "keyword": "duration", "enum": { "0": "0", "48": "48" } }, "next_port_id": "next_port_id" } }, "public-global-nitrate": { "name": "Konsentrasi nitrat(NO3)", "description": "Global Fishing Watch detects vessel movements in and out of ports and classifies them into four distinct port event types:
These events are grouped to form port visit events, which represent a vessel’s presence at port based on AIS activity patterns.
In the Global Fishing Watch map, only high-confidence port visits are shown. A port visit is determined with high confidence when a vessel is detected with a port entry, stop or gap, and exit. More specifically:
A port visit is shown on the map when AIS data shows a vessel:
Port stops are used to distinguish actual port visits from coastal transits.
Learn more about anchorages, ports and voyages.
Learn more about considerations of using AIS data by looking at the AIS limitations section in the Apparent fishing events (AIS) data documentation..", "schema": { "type": { "keyword": "type", "enum": { "CARGO": "CARGO", "DISCREPANCY": "DISCREPANCY", "CARRIER": "CARRIER", "FISHING": "FISHING", "GEAR": "GEAR", "OTHER": "OTHER", "PASSENGER": "PASSENGER", "SEISMIC_VESSEL": "SEISMIC_VESSEL", "SUPPORT": "SUPPORT" } }, "duration": { "keyword": "duration", "enum": { "2": "2", "48": "48" } } } }, "public-global-ports-footprint": { "name": "Anchorages convex hulls by portId", "description": "Anchorages footprint using the convex hull grouped by portId", "schema": { "area": "area", "label": { "keyword": "label", "enum": {} }, "perimeter": "perimeter" } }, "public-global-presence-tracks": { "name": "Tracks", "description": "The dataset contains the tracks from all vessels (AIS) - Version 20201001", "schema": { "speed": { "keyword": "speed", "enum": { "0": "0", "20": "20" } } } }, "public-global-presence": { "name": "AIS", "description": "Global Fishing Watch uses data about a vessel’s identity, type, location, speed, direction and more that is broadcast using the Automatic Identification System (AIS) and collected via satellites and terrestrial receivers. AIS was developed for safety/collision-avoidance. Global Fishing Watch analyzes AIS data collected from vessels that our research has identified as carriers. The activity layer displays a heatmap of vessel presence. The presence is determined by taking one position per day per vessel from the positions transmitted by the vessel's AIS.", "schema": { "speed": { "keyword": "speed", "enum": { "<2": "<2", "2-4": "2-4", "4-6": "4-6", "6-10": "6-10", "10-15": "10-15", "15-25": "15-25", ">25": ">25" } }, "bearing": "bearing" } }, "public-global-salinity": { "name": "Kadar garam", "description": "
Satellite synthetic aperture radar (SAR) is a spaceborne radar imaging system that can detect at-sea vessels and structures in any weather conditions. Microwave pulses are transmitted by a satellite-based antenna towards the Earth surface. The microwave energy scattered back to the spacecraft is then measured and integrated to form a “backscatter” image. The SAR image contains rich information about the different objects on the water, such as their size, orientation and texture. SAR imaging systems overcome most weather conditions and illumination levels, including clouds or rain due to the cloud penetrating property of microwaves, and daylight or darkness due to radar being an “active” sensor (it shoots and records back its own energy). SAR gives an advantage over some other “passive” satellite sensors, such as electro-optical imagery, consisting of a satellite-based camera recording the sunlight/infrared radiation reflected from/emitted by objects on the ground. This latter method can be confounded by cloud cover, haze, weather events and seasonal darkness at high latitudes.
\nWe use SAR imagery from the Copernicus Sentinel-1 mission of the European Space Agency (ESA) [1]. The images are sourced from two satellites (S1A and S1B up until December 2021 when S1B stopped operating, and S1A only from 2022 onward) that orbit 180 degrees out of phase with each other in a polar, sun-synchronous orbit. Each satellite has a repeat-cycle of 12 days, so that together they provide a global mapping of coastal waters around the world approximately every six days for the period that both were operating. The number of images per location, however, varies greatly depending on mission priorities, latitude, and degree of overlap between adjacent satellite passes. Spatial coverage also varies over time [2]. Our data consist of dual-polarization images (VH and VV) from the Interferometric Wide (IW) swath mode, with a resolution of about 20 m.
\n[1]\n \n https://sedas.satapps.org/wp-content/uploads/2015/07/Sentinel-1_User_Handbook.pdf\n \n
\n[2]\n \n https://sentinels.copernicus.eu/web/sentinel/missions/sentinel-1/observation-scenario\n \n
\nDetection footprints are areas within each satellite scan (or scene) that our system uses to perform detections. These filters help to keep relevant detections and exclude data that may be inaccurate. Detection footprints are smaller than the total scene as they exclude any land areas and islands, and exclude a 500 meter buffer from the boundaries of the scene and a 1 kilometer buffer from shorelines.
\nGFW has post-processed the SAR detections to reduce noise (false positives), remove offshore infrastructure from this layer focused on vessels, and exclude areas with sea ice at high latitudes.
\nDetecting vessels with SAR is based on an known as Constant False Alarm Rate (CFAR), a threshold algorithm used for anomaly detection in radar imagery. This algorithm is designed to search for pixel values that are unusually bright (the targets) compared to those in the surrounding area (the sea clutter). This method sets a threshold based on the pixel values of the local background (within a window), scanning the whole image pixel-by-pixel. Pixel values above the threshold constitute an anomaly and are likely to be samples from a target, and therefore are included as a detection.
\nTo estimate the length of every detected object and also to identify when our CFAR algorithm made false detections, we designed a deep convolutional neural network (ConvNet) based on the modern ResNet (Residual Networks) architecture. This single-input/multi-output ConvNet takes dual-band SAR image tiles of 80 by 80 pixels as input, and outputs the probability of object presence (known as a “binary classification task”) and the estimated length of the object (known as a “regression task”).
\nTo identify whether a detected vessel was a fishing or non-fishing vessel we use a machine learning model. For this classification task we used a ConvNeXt architecture modified to process the following two inputs: the estimated length of the vessel from SAR (a scalar quantity) and a stack of environmental rasters centered at the vessel’s location (a multi-channel image). This multi-input-mixed-data/single-output model passes the raster stack (11 channels) through a series of convolutional layers and combines the resulting feature maps with the vessel length value to perform a binary classification: fishing or non-fishing.
\nThe environmental layers used to differentiate between fishing and non-fishing include:
\nAIS data can reveal the identity of vessels, their owners and corporations, and fishing activity. Not all vessels, however, are required to use AIS devices, as regulations vary by country, vessel size, and activity. Vessels engaged in illicit activities can also turn off their AIS transponders or manipulate the locations they broadcast. Also, large “blind spots” along coastal waters arise from nations that restrict access to AIS data that are captured by terrestrial receptors instead of satellites or from poor reception due to high vessel density and low-quality AIS devices. Unmatched SAR detections therefore provide the missing information about vessel traffic in the ocean.
\nMatching SAR detections to vessels’ GPS coordinates (from the automatic identification system (AIS) is challenging because the timestamp of the SAR images and AIS records do not coincide, and a single AIS message can potentially match to multiple vessels appearing in the image, and vice versa. To determine the likelihood that a vessel broadcasting AIS corresponded to a specific SAR detection, we followed a matching approach based on probability rasters of where a vessel is likely to be minutes before and after an AIS position was recorded. These rasters were developed from one year of global AIS data from the Global Fishing Watch pipeline which uses Spire Global and Orbcomm sources of satellite data, including roughly 10 billion vessel positions, and computed for six different vessel classes, considering six different speeds and 36 time intervals. So we obtain the likely position of a vessel that could match a SAR detection based on the vessel class, speed and time interval.
\nAutomatic identification system (AIS) data can reveal the identity of vessels, their owners and corporations, and fishing activity. Not all vessels, however, are required to use AIS devices, as regulations vary by country, vessel size, and activity. Vessels engaged in illicit activities can also turn off their AIS transponders or manipulate the locations they broadcast. Also, large “blind spots” along coastal waters arise from nations that restrict access to AIS data that are captured by terrestrial receptors instead of satellites or from poor reception due to high vessel density and low-quality AIS devices. Unmatched SAR detections therefore provide the missing information about vessel traffic in the ocean.
\nAll code developed in this study for SAR detection, deep learning models, and analyses is open source and freely available at\n \n https://github.com/GlobalFishingWatch/paper-industrial-activity\n .\n
\nAll vessel data are freely available through the Global Fishing Watch data portal at\n \n https://globalfishingwatch.org\n . All data to reproduce our supporting scientific paper can be downloaded from\n \n https://doi.org/10.6084/m9.figshare.24309475\n \n (statistical analyses and figures) and\n \n https://doi.org/10.6084/m9.figshare.24309469\n \n (model training and evaluation).\n
\nNon-Commercial Use Only. The Site and the Services are provided for Non-Commercial use only in accordance with the CC BY-NC 4.0 license. If you would like to use the Site and/or the Services for commercial purposes, please contact us.", "schema": { "speed": "speed", "bearing": "bearing", "matched": { "keyword": "matched", "enum": { "true": "1", "false": "false" } }, "shiptype": { "keyword": "shiptype", "enum": { "carrier": "carrier", "seismic_vessel": "seismic_vessel", "passenger": "passenger", "other": "other", "support": "support", "bunker": "bunker", "gear": "Peralatan", "cargo": "cargo", "fishing": "fishing", "discrepancy": "discrepancy" } }, "neural_vessel_type": { "keyword": "neural_vessel_type", "enum": { "Likely non-fishing": "Likely non-fishing", "Likely Fishing": "Likely Fishing", "Unknown": "Tidak diketahui" } } } }, "public-global-sentinel2-footprints": { "name": "Detection footprints", "description": "Detection footprints" }, "public-global-sentinel2-presence": { "name": "Imagery detections (Optical)", "description": "
This layer shows vessels detected using optical satellite imagery collected by the European Space Agency's Sentinel-2 satellites. Optical imagery is similar to high-quality aerial photography from space, using reflected sunlight in visible and near-infrared wavelengths. This type of imagery provides high-resolution detail that allows us to spot small vessels, identify wake patterns, and better understand activity near shore.
Global Fishing Watch uses a machine learning model that processes each image to identify vessels and estimate their length, orientation, and speed based on wake features. The detections are then filtered using a secondary classifier to remove objects that are not vessels, such as clouds, rocks or icebergs. Each detection is linked to a cropped image (a thumbnail) so users can visually inspect what the model identified.
Because optical satellites rely on sunlight and clear skies, detections are only possible during the day and when the area is not obscured by clouds or haze. Despite these limitations, detections with optical imagery are especially helpful in identifying small untracked vessels that may not appear in other tracking systems.
This layer is based on images from the Sentinel-2 satellites operated by the European Space Agency (ESA). These satellites capture medium-resolution images (10 m per pixel) of the ocean using visible and near-infrared light (among several other bands). Combined, the satellites acquire images of most coastal waters and dedicated areas in the open ocean roughly every five days, and the imagery is made freely available by the ESA.
We use pre-processed Sentinel-2 images that have been corrected for geometric distortions and aligned to the Earth's surface. These images are split into manageable tiles, and we selected the tiles that cover only ocean areas (image tiles over land are excluded). We use four image bands: red, green, blue (RGB), and near-infrared (NIR), all at 10-meter resolution. These bands give us the detail and contrast needed to detect and classify vessels.
Our machine learning model scans each image tile to detect vessels. It is trained to look for features such as the shape, brightness, and wake of a vessel. When it finds a likely candidate, the model predicts a score for vessel presence alongside estimates of the vessel's location, size, orientation, and speed.
The detection model was trained on over 11,000 manually reviewed vessel examples across thousands of Sentinel-2 scenes. This training process included many small vessels and scenes from around the world, helping the model to perform well across different environments and vessel types.
Each detection includes a small visual \"chip\" showing the detected vessel at the center. These thumbnails come in two formats: a color version from the RGB bands, and a grayscale version from the near-infrared band. Each chip covers an area of 1 km². These thumbnails are helpful for visually confirming a detection or understanding its context. For very small vessels (under 15 meters), it may still be difficult to see them clearly.
Not everything that looks like a vessel in satellite imagery actually is one. To help remove false detections (like buoys, offshore platforms, sea ice, or clouds), we run each detection through a secondary classifier. This classifier is a machine learning model that uses both the image thumbnail and additional information about the detection (such as distance from shore, local depth, and vessel density nearby, among others) to decide whether the object is likely to be a vessel. We also flag detections that are close to known fixed infrastructure or in areas with substantial sea ice or iceberg presence.
If a detection is classified as likely non-vessel or flagged as potential infrastructure or ice, we remove it from the map layer so only high-confidence detections are included. We also clip the satellite footprints (displayed on the map layer) to exclude the areas under the icy-region mask. However, we provide all the false positives with labels through the data download portal for stakeholders who require a more complete dataset.
AIS data can reveal the identity of vessels, their owners and corporations, and fishing activity. Not all vessels, however, are required to use AIS devices, as regulations vary by country, vessel size, and activity. Vessels engaged in illicit activities can also turn off their AIS transponders or manipulate the locations they broadcast. Also, large \"blind spots\" along coastal waters arise from nations that restrict access to AIS data that are captured by terrestrial receptors instead of satellites or from poor reception due to high vessel density and low-quality AIS devices. Unmatched imagery detections therefore provide the missing information about vessel traffic in the ocean.
Matching imagery detections to vessels' GPS coordinates from AIS is challenging because the timestamps of the images and AIS records do not coincide, and a single AIS identity can potentially match to multiple vessels appearing in the image, and vice versa. To determine the likelihood that a vessel broadcasting AIS corresponded to a specific detection, we developed a matching approach based on probability rasters of where a vessel is likely to be minutes before and after an AIS position was recorded. These rasters were produced from one year of global AIS data from the Global Fishing Watch pipeline, which sources satellite data from Spire Global and Orbcomm. The probability rasters are based on roughly 10 billion vessel positions and are computed for six different vessel classes, considering six different speeds and 36 time intervals. So we obtain the likely position of a vessel that could match a detection based on the vessel class, speed and time interval. In addition to the spatiotemporal matching, we factor in the similarity between the model-inferred vessel length and the length from AIS identity data to avoid (likely incorrect) matches with large discrepancies in size, e.g., AIS of a tugboat and the detection of a large vessel behind it.
To help users understand where detections were possible, we show the detection \"footprints\" on the map. These polygons are the portions of the satellite images that cover the ocean and that were used for detection. Thus, if you see a footprint but no detections, it means no vessels were detected in that area. If there is no footprint, no image was processed for that location and time.
Our detection and matching system runs automatically each day. It checks for new Sentinel-2 images published to Google Cloud and processes those that meet our quality criteria. New detections are typically available within 1–2 days of the satellite capturing the image. The automated pipeline also re-checks any images published late to ensure any data gaps are filled.
All vessel data are freely available through the Global Fishing Watch data portal at https://globalfishingwatch.org/data-download/.
Non-Commercial Use Only. The Site and the Services are provided for Non-Commercial use only in accordance with the CC BY-NC 4.0 license. If you would like to use the Site and/or the Services for commercial purposes, please contact us.
", "schema": { "length": { "keyword": "length", "enum": { "<20": "<20", "20-60": "20-60", "60-100": "60-100", ">100": ">100" } }, "bearing": "bearing", "matched": { "keyword": "matched", "enum": { "true": "true", "false": "false" } }, "shiptype": { "keyword": "shiptype", "enum": { "carrier": "carrier", "seismic_vessel": "seismic_vessel", "passenger": "passenger", "other": "other", "support": "support", "bunker": "bunker", "gear": "gear", "cargo": "cargo", "fishing": "fishing", "discrepancy": "discrepancy" } } } }, "public-global-sst-anomalies-max": { "name": "Sea Surface Temperatures anomalies (Max)", "description": "Sea surface temperatures anomalies (Max)" }, "public-global-sst-anomalies-min": { "name": "Sea Surface Temperatures anomalies (Min)", "description": "Sea surface temperatures anomalies (Max)" }, "public-global-sst-anomalies": { "name": "Sea Surface Temperatures anomalies", "description": "Sea surface temperatures anomalies (Mean)" }, "public-global-sst-max": { "name": "Surface Temperature Max", "description": "Surface Temperature" }, "public-global-sst-min": { "name": "Surface Temperature Min", "description": "Surface Temperature" }, "public-global-sst": { "name": "Suhu permukaan laut", "description": "The United Nations Convention on the Law of the Sea describes the high seas as ‘all parts of the sea that are not included in the exclusive economic zone, in the territorial sea or in the internal waters of a State, or in the archipelagic waters of an archipelagic State.’ High Seas pockets are areas totally enclosed by EEZs. These pockets can be hard to distinguish from the multiple EEZ jurisdictions that surround them, thus, we have a layer that highlights them. Citation: Flanders Marine Institute (2024). Maritime Boundaries Geodatabase: High Seas, version 2. Available online at https://www.marineregions.org/.
" }, "public-indonesia-fishing-effort": { "name": "Indonesia VMS", "description": "VMS data for Indonesia is not currently available for the period from July 2020.\n\nVessel monitoring system (VMS) data provided by the Indonesian Government’s Ministry of Maritime Affairs and Fisheries. Data is collected using their VMS via satellites and terrestrial receivers, and contains a vessel identities, gear type, location, speed, direction and more. Global Fishing Watch analyzes this data using the same algorithms developed for automatic identification system (AIS) data to identify fishing activity and behaviors. The algorithm classifies each broadcast data point for these vessels as either apparently fishing or not fishing and shows the former on the Global Fishing Watch fishing activity heat map. VMS broadcasts data quite differently from AIS and may give different measures of completeness, accuracy and quality. Over time our algorithms will improve across all our broadcast data formats. Global Fishing Watch’s fishing presence algorithm for VMS, as for AIS, is a best effort to algorithmically identify “apparent fishing activity.” It is possible that some fishing activity is not identified, or that the heat map may show apparent fishing activity where fishing is not actually taking place. For these reasons, Global Fishing Watch qualifies the terms “fishing activity,” “fishing” or “fishing effort,” as “apparent,” rather than certain. Any/all Global Fishing Watch information about “apparent fishing activity” should be considered an estimate and must be relied upon solely at your own risk. Global Fishing Watch fishing presence algorithms are developed and tested using actual fishing event data collected by observers, combined with expert analysis of AIS vessel movement data resulting in the manual classification of thousands of known fishing events. Global Fishing Watch also collaborates extensively with academic researchers through our research program to share fishing activity classification data and automated classification techniques." }, "public-indonesia-fishing-vessels": { "name": "Indonesia VMS (Fishing Vessels)", "description": "Dataset for VMS Indonesia (Public)" }, "public-indonesia-pelagic-fishing-effort": { "name": "Indonesia Pelagic (Pilot)", "description": "Indonesia Pelagic Fishing Effort Public Data" }, "public-indonesia-pelagic-presence": { "name": "Indonesia Pelagic (Pilot)", "description": "Pelagic Presence" }, "public-indonesia-zebrax-presence": { "name": "Indonesia Zebrax", "description": "Lapisan Global Fishing Watch ini menggunakan data yang disediakan oleh Rare, Aruna dan AP2HI. Data dikumpulkan menggunakan perangkat ZebraX yang melacak lokasi dan kecepatan. Informasi yang ditampilkan mewakili keberadaan kapal.", "schema": { "speed": "speed", "bearing": "bearing" } }, "public-inshore-fishing-zone-1618837176535": { "name": "Inshore Fishing Zone", "description": "50 nm around the Tristan Archipelago and 40nm around Gough (8% of EEZ)" }, "public-isa-areas-contractor": { "name": "ISA Areas by Contractor", "description": "| Contractor ID | Preferred Contractor Name | Nationality |
|---|---|---|
| BGRPMN1 | Federal Institute for Geosciences and Natural Resources of Germany - PMN | Germany |
| BGRPMS1 | Federal Institute for Geosciences and Natural Resources of Germany - PMS | Germany |
| BMJPMN1 | Blue Minerals Jamaica Limited (BMJ) - PMN | Jamaica |
| BPHDCPMN1 | Beijing Pioneer Hi-Tech Development Corporation - PMN | China |
| BrazilCRFC1 | Companhia de Pesquisa de Recursos Minerais S.A. - CRFC | Brazil |
| CIICPMN1 | Cook Islands Investment Corporation - PMN | Cook Islands |
| CMMPMN1 | China Minmetals Corporation - PMN | China |
| COMRACRFC1 | China Ocean Mineral Resources Research and Development Association - CRFC | China |
| COMRAPMN1 | China Ocean Mineral Resources Research and Development Association - PMN | China |
| COMRAPMS1 | China Ocean Mineral Resources Research and Development Association - PMS | China |
| DORDPMN1 | Deep Ocean Resources Development Co. Ltd. - PMN | Japan |
| GSRPMN1 | Global Sea Mineral Resources NV - PMN | Belgium |
| IFREMERPMN1 | Institut français de recherche pour l'exploitation de la mer - PMN | France |
| IFREMERPMS1 | Institut français de recherche pour l'exploitation de la mer - PMS | France |
| IndiaPMN1 | Government of India - PMN | India |
| IndiaPMS1 | Government of India - PMS | India |
| IOMPMN1 | Interoceanmetal Joint Organization - PMN | Poland |
| KOREACRFC1 | Government of the Republic of Korea - CRFC | Korea |
| KOREAPMN1 | Government of the Republic of Korea - PMN | Korea |
| KOREAPMS1 | Government of the Republic of Korea - PMS | Korea |
| MARAWAPMN1 | Marawa Research and Exploration Ltd. - PMN | Kiribati |
| NORIPMN1 | Nauru Ocean Resources Inc. - PMN | Nauru |
| OMSPMN1 | Ocean Mineral Singapore Pte. Ltd. - PMN | Singapore |
| POLPMS1 | Government of the Republic of Poland - PMS | Poland |
| RUSFEDPMS1 | Government of the Russian Federation - PMS | Russia |
| RUSMNRCRFC1 | Ministry of Natural resources and environment of the russian federation - CRFC | Russia |
| TOMLPMN1 | Tonga Offshore Mining Limited - PMN | Tonga |
| UKSRLPMN1 | UK Seabed Resources Ltd. - I - PMN | UK |
| UKSRLPMN2 | UK Seabed Resources Ltd. - II - PMN | UK |
| YUZHPMN1 | Yuzhmorgeologiya - PMN | Russia |
The MPA Guide is intended to fill a gap in existing MPA classification and assessment tools to help determine how likely MPAs are to deliver the desired conservation outcomes. Over the past few years, the MPAtlas team and key collaborators have applied The MPA Guide framework to MPAs around the world to develop a clearer picture of global marine protection, which includes a recent study establishing a baseline for global \"30x30\" targets - Publication: Ocean Protection Quality is Lagging Behind Quantity. MPAtlas is the online repository for MPA Guide assessments and associated data. Stage of Establishment represents where the MPA is in its process of being an MPA. Proposed/Committed: The intent to create an MPA is made public. Designated: MPA is established/recognized through legal means or other authoritative rulemaking. Implemented: MPA is acknowledged to be operation ‘in the water’ with plans for management activated. Actively managed: MPA management is ongoing, with monitoring, periodic review and adjustments made as needed to achieve biodiversity conservation and other ecological and social goals. Level of Protection represents the extent to which the MPA protected from seven main types of human activities and is likely to generate positive biodiversity outcomes. Fully Protected: No impact from extractive or destructive activities is allowed, and all abatable impacts are minimized. Highly Protected: Only light extractive activities are allowed that have low total impact, and all other abatable impacts are minimized. Lightly Protected: Some protection of biodiversity exists, but extractive or destructive activities that can have moderate to significant impact are allowed. Minimally Protected: Extensive extraction and other activities with high total impact are allowed, but the site can still be considered an MPA under the IUCN protected area definition and provides some conservation benefit. Some areas allow activities that have an impact so large that they are incompatible with the conservation of biodiversity, as defined by the IUCN. For more information, please visit https://mpatlas.org/. Each assessed MPA has a score card that describes its stage of establishment and level of protection, as well as more details about the components that contributed to these assessments. Note: The MPAtlas dataset does not contain boundaries for all global MPAs, only those assessed against MPA quality frameworks.
", "schema": { "establishment_stage": { "keyword": "establishment_stage", "enum": { "proposed/committed": "proposed/committed", "designated": "designated", "actively managed": "actively managed", "implemented": "implemented", "unknown": "unknown" } }, "mpaguide_protection_level": { "keyword": "mpaguide_protection_level", "enum": { "unknown": "unknown", "light": "light", "full": "full", "high": "high", "incompatible": "incompatible", "minimal": "minimal" } } } }, "public-mpa-all": { "name": "KKP (WDPA)", "description": "Marine protected areas (MPAs) are areas of the ocean set aside for long-term conservation. These can have different levels of protection, and the range of activities allowed or prohibited within their boundaries varies considerably. Source: World Database on Protected Areas. Last updated: November 2022. See more detailed metadata information for this layer." }, "public-mpa-no-take": { "name": "MPAs - No take", "description": "MPA berarti sebagai kawasan perlindungan laut. Kawasan lindung laut larang tangkap adalah kawasan lautan yang dilarang mengekstraksi atau merusak secara signifikan sumber daya alam dan budaya, termasuk ikan.Sumber: World Database on Protected Areas. Terakhir diperbarui: May 2021." }, "public-mpa-no-take-partial": { "name": "MPAs - No take Partial", "description": "The term Marine Protected Areas include marine reserves, fully protected marine areas, no-take zones, marine sanctuaries, ocean sanctuaries, marine parks, locally managed marine areas, to name a few. Many of these have quite different levels of protection, and the range of activities allowed or prohibited within their boundaries varies considerably too. Source: World Database on Protected Areas (WDPA)" }, "public-mpa-restricted": { "name": "MPAs - Restricted (Source: WDPA)", "description": "MPA berarti sebagai kawasan perlindungan laut. Penggunaan yang dibatasi dapat memiliki tingkat perlindungan yang berbeda, dan rentang kegiatan yang diizinkan atau dilarang di dalam batas-batasnya sangat bervariasi. Sumber: World Database on Protected Areas. Terakhir diperbarui: May 2021." }, "public-no-take-zone-1618836692786": { "name": "Marine Protection Zone", "description": "90% of EEZ" }, "public-norway-fishing-effort": { "name": "Norway VMS", "description": "Vessel monitoring system (VMS) data is provided by the The Norwegian Directorate of Fisheries. Data is collected using Norway’s vessel monitoring system via satellites and is published on a three-day delay containing information on vessels’ location, speed, course, and movement. Global Fishing Watch analyzes this data using the same algorithms developed for automatic identification system (AIS) to identify fishing activity and behaviors. The algorithm classifies each broadcast data point from vessels as either apparently fishing or not fishing and shows the former on the Global Fishing Watch’s fishing activity heat map. VMS broadcasts data differently from AIS and may give different measures of completeness, accuracy, and quality. Global Fishing Watch is continually improving its algorithms across all broadcast data formats to algorithmically identify “apparent fishing activity.” It is possible that some fishing activity is not identified or that the heat map may show apparent fishing activity when fishing is not actually taking place. For these reasons, Global Fishing Watch qualifies the terms “fishing activity,” “fishing” or “fishing effort,” as apparent rather than certain. Any and all Global Fishing Watch information about “apparent fishing activity” should be considered an estimate and must be relied upon solely at the user’s discretion. Global Fishing Watch’s fishing detection algorithms are developed and tested using actual fishing event data collected by observers and is combined with expert analysis of AIS vessel movement data, resulting in the manual classification of thousands of known fishing events. Global Fishing Watch also collaborates extensively with academic researchers through our research program to share fishing activity classification data and to improve automated classification techniques", "schema": { "speed": "speed", "bearing": "bearing" } }, "public-norway-fishing-vessels": { "name": "Norway VMS (Fishing Vessels)", "description": "Dataset for VMS Norway (Public)", "schema": { "source": "source" } }, "public-norway-non-fishing-vessels": { "name": "Norway VMS (Non Fishing Vessels)", "description": "Dataset for VMS Norway (Public)", "schema": { "source": "source" } }, "public-norway-presence": { "name": "Norway VMS", "description": "Dataset for VMS Norway presence", "schema": { "speed": "speed", "bearing": "bearing" } }, "public-norway-vessel-identity-fishing": { "name": "Norway VMS (Fishing Vessels)", "description": "Dataset for VMS Norway (Public)" }, "public-norway-vessel-identity-non-fishing": { "name": "Norway VMS (Non Fishing Vessels)", "description": "Dataset for VMS Norway (Public)" }, "public-paa-duke": { "name": "PAAS", "description": "This global Preferential Access Area (PAA) database was built with data for 44 countries identified by the Illuminating Hidden Harvests (IHH) initiative following Basurto et al. (2023) and the FAOLEX database (https://www.fao.org/faolex/en/). The total resulting PAAs identified in this study (n = 63) in 44 countries may be considered a lower bound and not comprehensive, as governments can also establish PAAs via laws or regulations either: (i) not relating to the fisheries sector and categorized elsewhere in the FAOLEX database and hence not searched, or (ii) not reported or captured in it. The most recent review of FAOLEX to identify PAAs was conducted in 2022. For mapping and spatial analysis of the PAAs, a geodatabase of PAA boundaries was created using definition parameters for each PAA (i.e., distance from shoreline or the baseline, depth, or a given set of coordinates or buffered distances around undersea features). Baselines were extracted from Flanders Marine Institute’s (2023) version 12 global EEZ database, defined by ‘Straight Baseline’ features. The shoreline dataset was ESRI’s 2014 global shoreline, the same shoreline dataset as the Flanders Marine Institute global EEZ database. Depth data was extracted from the GEBCO 2023 global relief dataset. When possible, Duke reached out to local experts to review the boundaries for confirmation that our interpretation of FAOLEX and the relevant law or regulation matches what is being used by local fishers. If you have feedback on the PAA boundaries, information regarding PAAs that are not reflected here, or other questions, please reach out to ssf-paa-info@duke.edu. For more information on PAAs and how they are managed, please see Duke’s publication by Basurto et al. (2024) here: https://doi.org/10.1038/s44183-024-00096-0. Please note that there is no singular definition of a PAA, therefore each country will decide what activities, vessels, gear types, etc. are prohibited or restricted in the PAA. For more information of the specific designation of each PAA, please see the law or regulation in FAOLex. Citation: DeLand, S., Vegh, T., Cleary, J., Basurto, X., Virdin, J., & Halpin, P. N. (2025). A global dataset of preferential access areas for small-scale fishing. Duke Research Data Repository. https://doi.org/10.7924/r40s01h5j
" }, "public-panama-fishing-effort": { "name": "Panama VMS", "description": "Vessel monitoring system (VMS) data provided by the Panamanian Authority of Aquatic Resources (ARAP). Data is received by Panama’s VMS system via satellite and contains vessel identities, gear type, location, speed, direction and more. Panama’s carrier vessel data is also available here. Each point in the carrier vessel data layer represents a position of the carriers, but not all positions are displayed. Carrier vessel positions are displayed once per day. In the future, we expect to be able to display more positions. Click on a carrier vessel’s position to view the vessel’s complete track. Global Fishing Watch analyzes this data using the same algorithms we developed for automatic identification system (AIS) data to identify fishing activity and behaviors. The algorithm classifies each broadcast data point from vessels as either apparently fishing or not fishing and shows the former on the Global Fishing Watch fishing activity heat map. VMS broadcasts data quite differently from AIS and may give different measures of completeness, accuracy and quality. Over time our algorithms will improve across all our broadcast data formats. Global Fishing Watch’s fishing presence algorithm for VMS, as for AIS, is a best effort to algorithmically identify “apparent fishing activity.” It is possible that some fishing activity is not identified, or that the heat map may show apparent fishing activity where fishing is not actually taking place. For these reasons, Global Fishing Watch qualifies the terms “fishing activity,” “fishing,” and “fishing effort,” as “apparent” rather than certain. Any/all Global Fishing Watch information about “apparent fishing activity” should be considered an estimate and must be relied upon solely at your own risk. Global Fishing Watch fishing presence algorithms are developed and tested using actual fishing event data collected by observers, combined with expert analysis of AIS vessel movement data resulting in the manual classification of thousands of known fishing events. Global Fishing Watch also collaborates extensively with academic researchers through our research program to share fishing activity classification data and automated classification.", "schema": { "speed": "speed", "bearing": "bearing" } }, "public-panama-fishing-vessels": { "name": "Panama VMS (Public Fishing Vessels)", "description": "Dataset for VMS Panama (Public)", "schema": { "source": "source" } }, "public-panama-non-fishing-vessels": { "name": "Panama VMS (Public Non fishing vessels)", "description": "Dataset for VMS Panama - Carriers (Public)", "schema": { "source": "source" } }, "public-panama-vessel-identity-fishing": { "name": "Panama VMS (Public Fishing Vessels)", "description": "Dataset for VMS Panama (Public)" }, "public-panama-vessel-identity-non-fishing": { "name": "Panama VMS (Public Non fishing vessels)", "description": "Dataset for VMS Panama - Carriers (Public)" }, "public-peru-fishing-effort": { "name": "Peru VMS", "description": "Vessel monitoring system (VMS) data provided by the Peruvian Government’s Ministry of Production, Fisheries Sector (PRODUCE). Permission to include Peruvian Data required that a 10 day delay to publishing was implemented. Data is collected using their vessel monitoring system (VMS) via satellites and terrestrial receivers, and contains a vessel’s identity gear type, location, speed, direction and more. Global Fishing Watch analyzes this data using the same algorithms developed for automatic identification system (AIS) data to identify fishing activity and behaviors. The algorithm classifies each broadcast data point for these vessels as either apparently fishing or not fishing and shows the former on the Global Fishing Watch fishing activity heat map. VMS broadcasts data quite differently from AIS and may give different measures of completeness, accuracy and quality. Over time our algorithms will improve across all our broadcast data formats. Global Fishing Watch’s fishing presence algorithm for VMS, as for AIS, is a best effort to algorithmically identify “apparent fishing activity.” It is possible that some fishing activity is not identified, or that the heat map may show apparent fishing activity where fishing is not actually taking place. For these reasons, Global Fishing Watch qualifies the terms “fishing activity,” “fishing” or “fishing effort,” as “apparent,” rather than certain. Any/all Global Fishing Watch information about “apparent fishing activity” should be considered an estimate and must be relied upon solely at your own risk. Global Fishing Watch fishing presence algorithms are developed and tested using actual fishing event data collected by observers, combined with expert analysis of AIS vessel movement data resulting in the manual classification of thousands of known fishing events. Global Fishing Watch also collaborates extensively with academic researchers through our research program to share fishing activity classification data and automated classification techniques.", "schema": { "fleet": { "keyword": "fleet", "enum": { "industrial": "industrial", "artisanal": "artisanal", "small-scale": "small-scale", "not defined": "not defined" } }, "speed": "speed", "origin": { "keyword": "origin", "enum": { "PER": "PER", "Foreign": "Foreign" } }, "bearing": "bearing" } }, "public-peru-fishing-vessels": { "name": "Peru VMS (Fishing Vessels)", "description": "Dataset for VMS Peru (Public)" }, "public-peru-presence": { "name": "Peru VMS Presence", "description": "Vessel monitoring system (VMS) data provided by the Peruvian Government’s Ministry of Production, Fisheries Sector (PRODUCE). Permission to include Peruvian Data required that a 10 day delay to publishing was implemented. Data is collected using their vessel monitoring system (VMS) via satellites and terrestrial receivers, and contains a vessel’s identity gear type, location, speed, direction and more. Global Fishing Watch analyzes this data using the same algorithms developed for automatic identification system (AIS) data to identify fishing activity and behaviors. The algorithm classifies each broadcast data point for these vessels as either apparently fishing or not fishing and shows the former on the Global Fishing Watch fishing activity heat map. VMS broadcasts data quite differently from AIS and may give different measures of completeness, accuracy and quality. Over time our algorithms will improve across all our broadcast data formats. Global Fishing Watch’s fishing detection algorithm for VMS, as for AIS, is a best effort to algorithmically identify “apparent fishing activity.” It is possible that some fishing activity is not identified, or that the heat map may show apparent fishing activity where fishing is not actually taking place. For these reasons, Global Fishing Watch qualifies the terms “fishing activity,” “fishing” or “fishing effort,” as “apparent,” rather than certain. Any/all Global Fishing Watch information about “apparent fishing activity” should be considered an estimate and must be relied upon solely at your own risk. Global Fishing Watch fishing detection algorithms are developed and tested using actual fishing event data collected by observers, combined with expert analysis of AIS vessel movement data resulting in the manual classification of thousands of known fishing events. Global Fishing Watch also collaborates extensively with academic researchers through our research program to share fishing activity classification data and automated classification techniques.", "schema": { "fleet": { "keyword": "fleet", "enum": { "industrial": "industrial", "artisanal": "artisanal", "small-scale": "small-scale", "not defined": "not defined" } }, "origin": { "keyword": "origin", "enum": { "PER": "PER", "Foreign": "Foreign" } } } }, "public-peru-vessel-identity-fishing": { "name": "Peru VMS (Fishing Vessels)", "description": "Dataset for VMS Peru (Public)" }, "public-png-fishing-effort": { "name": "Papua New Guinea VMS", "description": "Vessel monitoring system (VMS) data is provided by the The National Fisheries Authority of Papua New Guinea. Data is collected using Papua New Guinea's vessel monitoring (VMS) system via satellites, that contains vessel's identifiers and location, and is published on a five-day delay. Global Fishing Watch infers speed and course for each vessel location and analyzes this data using the same algorithms developed for automatic identification system (AIS) to identify fishing activity and behaviors. The algorithm classifies each broadcast data point from vessels as either apparently fishing or not fishing and shows the former on the Global Fishing Watch’s fishing activity heat map. VMS broadcasts data differently from AIS and may give different measures of completeness, accuracy, and quality. Global Fishing Watch is continually improving its algorithms across all broadcast data formats to algorithmically identify “apparent fishing activity”. It is possible that some fishing activity is not identified or that the heat map may show apparent fishing activity when fishing is not actually taking place. For these reasons, Global Fishing Watch qualifies the terms “fishing activity”, “fishing” or “fishing effort”, as apparent rather than certain. Any and all Global Fishing Watch information about “apparent fishing activity” should be considered an estimate and must be relied upon solely at the user’s discretion. Global Fishing Watch’s fishing detection algorithms are developed and tested using actual fishing event data collected by observers and is combined with expert analysis of AIS vessel movement data, resulting in the manual classification of thousands of known fishing events. Global Fishing Watch also collaborates extensively with academic researchers through our research program to share fishing activity classification data and to improve automated classification techniques", "schema": { "speed": "speed", "bearing": "bearing" } }, "public-png-fishing-vessels": { "name": "Papua New Guinea VMS (Fishing Vessels)", "description": "Dataset for VMS Papua New Guinea (Public)", "schema": { "source": "source" } }, "public-png-presence": { "name": "Papua New Guinea VMS", "description": "Vessel monitoring system (VMS) data is provided by the The National Fisheries Authority of Papua New Guinea. Data is collected using Papua New Guinea's national VMS that is provided by the Fisheries Information and Management System (FIMS). VMS data includes vessel identifiers and location, and is published with a five-day delay.\n\nThe activity layer displays a heatmap of vessel presence. The presence is determined by taking two positions per hour per vessel from the positions transmitted by the vessel's VMS.", "schema": { "speed": "speed", "bearing": "bearing" } }, "public-png-vessel-identity-fishing": { "name": "Papua New Guinea VMS (Fishing Vessels)", "description": "Dataset for VMS Papua New Guinea (Public)" }, "public-ports": { "name": "Ports", "description": "Named ports", "schema": { "name": "name", "port_id": "port_id" } }, "public-ports-v1": { "name": "Ports v1", "description": "Named ports v1" }, "public-presence-viirs-match-prototype": { "name": "VIIRS Match", "description": "The night lights vessel detections layer, known as visible infrared imaging radiometer suite or VIIRS, shows vessels at sea that satellites have detected by the light they emit at night. Though not exclusively associated with fishing vessels, this activity layer is likely to show vessels associated with activities like squid fishing, which use bright lights and fish at night.The satellite makes a single over-pass across the entire planet every night, detecting lights not obscured by clouds and designed to give at least one observation globally every day. Because the vessels are detected solely based on light emission, we can detect individual vessels and even entire fishing fleets that are not broadcasting automatic identification system (AIS) and so are not represented in the AIS apparent fishing effort layer. Lights from fixed offshore infrastructure and other non-vessel sources are excluded. Global Fishing Watch ingests boat detections processed from low light imaging data collected by the U.S. National Oceanic and Atmospheric Administration (NOAA) VIIRS. The boat detections are processed in near-real time by NOAA’s Earth Observation Group, located in Boulder, Colorado. The data, known as VIIRS boat detections, picks up the presence of fishing vessels using lights to attract catch or to conduct operations at night. More than 85% of the detections are from vessels that lack AIS or Vessel Monitoring System (VMS) transponders. Due to the orbit design of polar orbiting satellites, regions closer to polar will have more over-passes per day, while equatorial regions have only one over-pass daily. Read more about this product, and download the data here.Those using night light detections data should acknowledge the South Atlantic Anomaly (SAA), an area where the Earth's inner Van Allen radiation belt is at its lowest altitude, allowing more energetic particles from space to penetrate. When such particles hit the sensors on a satellite, this can create a false signal which might cause the algorithm to recognize it as a boat presence. A filtration algorithm has been applied but there may still be some mis-identification. The GFW layer includes quality flags (QF), including a filter to show only detections which NOAA has classified as vessels (QF1)", "schema": { "source": { "keyword": "source", "enum": { "unknown": "unknown", "chile_vms_aquaculture": "chile_vms_aquaculture", "peru.trasat_api": "peru.trasat_api", "namibia_vms": "namibia_vms", "indovms": "indovms", "mexico_vms": "mexico_vms", "chile_vms_small_fisheries": "perikanan kecil", "peru.historic": "peru.historic", "chile_vms_industry": "industri", "ais_scored": "ais_scored", "AIS": "AIS", "panama_vms": "panama_vms", "chile_vms_transport": "chile_vms_transport" } }, "matched": { "keyword": "matched", "enum": { "true": "1", "false": "false" } }, "radiance": { "keyword": "radiance", "enum": { "1": "1", "1000": "1000" } }, "shiptype": { "keyword": "shiptype", "enum": { "unknown": "unknown", "fishing": "fishing", "carrier": "carrier", "support": "support" } }, "qf_detect": { "keyword": "qf_detect", "enum": { "1": "1", "2": "2", "3": "3", "5": "5", "7": "7", "10": "10" } } } }, "public-protectedseas": { "name": "KKP (ProtectedSeas)", "description": "Marine Protected Areas (MPA) adalah kawasan laut yang dilindungi untuk konservasi jangka panjang. Kawasan Ini dapat memiliki tingkat perlindungan yang berbeda, dan rentang kegiatan yang diperbolehkan atau dilarang dalam batas-batasnya bervariasi. Sumber: ProtectedSeas Navigator. ProtectedSeas menetapkan skor Level of Fishing Protection (LFP) untuk setiap area berdasarkan analisis pembatasan ekstraksi biota laut. Anda dapat menggunakan ikon filter untuk menjelajahi skor LFP yang berbeda, yaitu, Tidak ada batasan: Tidak ada batasan untuk kegiatan penangkapan i; Kurang membatasi: Beberapa pembatasan khusus spesies atau alat tangkap berlaku; Cukup membatasi: Beberapa pembatasan khusus spesies atau alat tangkap berlaku; Sangat membatasi: Penangkapan ikan sebagian besar dilarang, dengan sedikit pengecualian; Paling membatasi: Pelarangan kegiatan penangkapan ikan. Lihat informasi lebih rinci untuk lapisan ini.", "schema": { "removal_of": { "keyword": "Tingkat Perlindungan Penangkapan Ikan", "enum": { "1": "1. Tidak ada batasan ", "2": "2. Kurang membatasi", "3": "3. Cukup membatasi", "4": "4. Sangat membatasi", "5": "5. Paling membatasi" } } } }, "public-protectedseas-all": { "name": "Protected Seas all", "description": "Protected seas all", "schema": { "category_name": { "keyword": "category_name", "enum": { "IUCN MPA": "IUCN MPA", "Other": "Lainnya", "Jurisdictional Authority Area": "Jurisdictional Authority Area", "TBD": "TBD", "Fisheries Management Area": "Fisheries Management Area", "Recreational Area": "Recreational Area", "OECM": "OECM", "Vessel Restricted Area": "Vessel Restricted Area", "Voluntary Conservation Measure Area": "Voluntary Conservation Measure Area", "Vessel Reporting Area": "Vessel Reporting Area", "Water Quality/Human Health Area": "Water Quality/Human Health Area" } }, "removal_of_marine_life_is_prohibited": { "keyword": "removal_of_marine_life_is_prohibited", "enum": { "1": "1. Tidak ada batasan ", "2": "2. Kurang membatasi", "3": "3. Cukup membatasi", "4": "4. Sangat membatasi", "5": "5. Paling membatasi" } } } }, "public-rfmo": { "name": "RFMO", "description": "Regional fisheries management organizations (RFMOs) are international bodies formed by countries with a shared interest in managing or conserving fish stocks in a particular region. Some manage all the fish stocks found in a given area, while others focus on specific highly migratory species, notably tuna. The regional fisheries management organization on the Global Fishing Watch map currently includes the five tuna regional fisheries management organizations. See more detailed metadata information for this layer.", "schema": { "ID": { "keyword": "ID", "enum": { "APFIC": "APFIC", "BOBP-IGO": "BOBP-IGO", "CCAMLR": "CCAMLR", "CCBSP": "CCBSP", "CCSBT": "CCSBT", "CCSBT Primary Area": "CCSBT Primary Area", "COREP": "COREP", "CPPS": "CPPS", "CRFM": "CRFM", "CTMFM": "CTMFM", "FCWC": "FCWC", "FFA": "FFA", "GFCM": "GFCM", "IATTC": "IATTC", "ICCAT": "ICCAT", "ICES": "ICES", "IOTC": "IOTC", "IPHC": "IPHC", "LTA": "LTA", "NAFO": "NAFO", "NAMMCO": "NAMMCO", "NASCO": "NASCO", "NEAFC": "NEAFC", "NPAFC": "NPAFC", "NPFC": "NPFC", "OSPESCA": "OSPESCA", "PERSGA": "PERSGA", "PICES": "PICES", "RECOFI": "RECOFI", "SEAFDEC": "SEAFDEC", "SIOFA": "SIOFA", "SPC": "SPC", "SPRFMO": "SPRFMO", "SRFC": "SRFC", "SWIOFC": "SWIOFC", "WCPFC": "WCPFC" } } } }, "public-seagrasses": { "name": "Lamun", "description": "