Start Point to Plymouth Sound and Eddystone

Intelligence Report – July 2022

The area is a good example of the complexities of managing an MPA that crosses various maritime boundaries.

General Key Finding

The use of vessel tracking combined with machine learning algorithms benefit the Special Area of Conservation the most.

Vessel Tracking Key Finding

Machine Learning identified 33 suspected non-compliant activities within byelaw areas in the Special Area of Conservation.

Remote Sensing Key Finding

There are low numbers of ‘dark vessels’ operating within the Special Area of Conservation. Further checks on VMS would confirm this.

Observed Activity

AIS Analysis Alerts:10 (3 Vessels)
Machine Learning Alerts:33 (20 vessels)
High Risks:0


Machine Learning is a very useful tool when applied to iVMS and VMS data, it can highlight when vessels are likely to be fishing in prohibited zones. There were a greater number of ‘dark’ vessels in the size range of possible fishing vessels in the offshore areas of the site.

Electro-optical imagery was an effective tool for monitoring activity in the Special Area of Conservation but had limited availability during the monitoring period. Additional imagery could provide further intelligence over the offshore sites.


Continue using Machine Learning to generate alerts over the UK MPA’s and pass the arising intelligence to the relevant authorities.

Continue to task electro-optical imagery over the site. This information could potentially contribute to MCS intelligence ahead of planned patrol efforts.

The image to the right shows ‘dark vessel’ detections in and around the Eddystone, Plymouth Sound and Start Point Special Area of Conservation.

Each orange marker denotes a ‘dark vessel’ detection from SAR (n=9) and electro-optical (n=3) images taken between May and July 2022.
Low risk detections were dispersed in the offshore areas of the site. This presents a logistical challenge for patrol efforts, monitoring, and surveillance for the competent authorities responsible for management of the site.

This low number and distribution of potential fishing vessels detections supports the utilization of a ‘Beyond Visual Line of Sight’ drone to carry out surveillance in the offshore areas of the site. These flights could represent a safe, cost-effective alternative to traditional patrol vessel deployments.

©2022 Microsoft, ©2022 Natural Earth, ©2022 exactEarth, ©2022 Flanders Marine Institute, ©2022 Spire, ©2022MDA, ©2022 MAXAR Technologies

Extended Findings

There were no high-risk SAR or electro-optical detections within the MPA, with 12 detections being categorised as low risk. These were likely to be fishing vessels, however based on the imagery we were unable to verify activity. The majority of detections were in the offshore proportion of the site, where patrol effort is more costly and time intensive to provide. It is possible that these vessels are transmitting on VMS, and it is recommended to carry out further checks to confirm this.

The availability of high-resolution imagery was limited over the site during the monitoring period. Existing imagery indicated a high level of vessel traffic (pleasure traffic and merchant traffic rather than fishing vessels), particularly in the inshore sector of the site.

Synthetic Aperture Radar (SAR) provided insights in the levels of activity within the area but is limited in its support for the monitoring, control and surveillance (MCS) of the UK MPAs. This is because of the diverse nature of vessel types and the time taken for image delivery and analysis and the nature of fishing activity in the area.

Risk Assessment

The primary risk to this site is from vessels illegally fishing with bottom towed gear within the Eddystone and the D&S IFCA byelaw areas. Other risks include breaches of more technical or gear related restrictions, that cannot be detected using the current suite of remote sensing tools available. However, the Machine Learning (ML) algorithm could be retrained to incorporate more detailed gear types within the analysis to detect potential high-risk activity.

Next Steps

Based on information from remote sensing sources there is a risk of unauthorised vessel activity over the MPA, this is shown by the number of SAR and EO detections and supported by the ML alerts of likely fishing activity.

Based on the key risks identified and information gathered in the project, it is recommended to continue monitoring the area using ML and generating alerts of likely fishing activity over byelaw areas for further investigation. This could be expanded with the national roll out of iVMS. This would mean that the area over Eddystone could be more comprehensively monitored. Alerts generated using data from the iVMS pilot show that ML is highly effective when analysing iVMS data.

Furthermore, it is recommended to effectively communicate publicly the use of electro-optical imagery and machine learning as this could offer a useful deterrent effect, which is crucial for overall compliance and effective MPA management.