top of page

Maritime Custody Service (MCS) Case Study: A BlackSky & Spire Global Partnership

The Maritime Custody Service (MCS) is a newly developed collaboration between BlackSky and Spire Global that will use an AI-powered vessel surveillance system designed for robust on-demand monitoring of the maritime domain.

Ships at a seaside port near a city

With the massive increase in maritime traffic and global trade throughout the last century, the need for enhanced maritime domain awareness is more apparent than ever. Everything from global trade to commercial fishing is putting a strain on the world’s environment, resources, and operations, and new technologies are opening doors for bad actors to conduct illicit and illegal activity unnoticed. 


To curb the impacts of these threats, maritime authorities need access to innovative technologies and strategies that provide real-time insights and offer pathways to solutions. A big part of the solution is to provide authorities with actionable, data-backed insights into vessel identification and monitoring, which help manage issues like illicit Ship-to-Ship (STS) cargo transfers, IUU fishing, trafficking, and sanctions compliance


By understanding the impacts of how shipping organizations and individual vessels operate at sea, including the impacts of shipping delays and port congestion, solutions can be strategized to ensure national security and provide resilience in trade compliance throughout global value chains. 


Until recently, however, maritime surveillance has lacked the capabilities to develop efficient and cost-effective strategies to mitigate some of the biggest threats to the maritime domain. These insufficient capabilities and methods include the sole use of Automatic Identification Systems (AIS) for vessel tracking that most ships are equipped with, as operators can simply shut off the system and stop tracking their locations and movements. Still, AIS is highly-valuable when used concurrently with other earth observation (EO) technologies like Synthetic Aperture Radar (SAR), Radio Frequency (RF) emissions, and satellite imagery. 


When applied correctly, strategies using multiple earth observation technologies can accurately and affordably identify, track, and monitor vessels at sea - leading to enhanced abilities to classify vessels as potential threats and mitigate related issues in the future. One such method is called ‘Tip and Cueing’ - a strategy that uses two or more satellites with varying degrees of sensors to monitor and track vessels or areas of interest. 

Maritime Custody Service (MCS) Overview


The Maritime Custody Service (MCS) is a newly developed collaboration between BlackSky and Spire Global that will use an AI-powered vessel surveillance system designed for robust on-demand monitoring of the maritime domain. The MCS is powered by satellite data collected by BlackSky and Spire Global constellations and uses BlackSky’s cloud-based imagery and analytics software (Spectra AI). 


Spectra AI is an automated, low-latency system that supports satellite imaging, data processing, and analytics using 26 EO satellites and can incorporate and analyze various external data sources for a more accurate overview of objects or locations of interest. The system will utilize its unique capabilities to continuously monitor vessels while docked at seaports and as they transit through open oceans, rivers, and canals. 


Below, we look at how this process works, where the service currently stands, and what can be done to improve the surveillance process. 

Concept of Operations (ConOps)


The MCS operates cyclically to maintain custody of vessels. It uses RF-based vessel tracking to identify a target RF signature, predict a vessel's trajectory with prediction modeling, identify potential imaging opportunities, ‘tip’ BlackSky’s satellites to execute imaging, analyze the imagery for detection, and update a target's geolocation. 


This ‘Tip and Cue’ approach allows for continuous monitoring of vessels by collecting and analyzing different electromagnetic spectrum wavelengths - ultimately providing actionable insights at every stage of the process. 


The process is outlined step-by-step in detail below. 


1. Identify RF Geolocation Target: Using RF emissions and AIS data collected by Spire satellites and AIS network, the MCS identifies the current positioning of a vessel, as well as its speed and heading.


​2. Vessel Trajectory Prediction: Once a vessel’s data is identified, the MCS predicts its trajectory by analyzing its position and velocity information collected in phase 1. 


  • As of now, the MCS uses a ‘dead-reckoning’ approach, where the trajectory of a vessel is assessed using its speed and direction. The simplified dead-reckoning trajectory currently uses straight lines for the predicted movements, but BlackSky has now developed a more advanced model that will anticipate future positioning in more detail (outlined in the ‘Results and Improvements’ section below). 


3. Image Opportunity Identification: From there, BlackSky identifies viable imaging opportunities by intersecting the ship’s probable trajectory. 


  • Opportunities that are most likely to intercept are integrated into a mission plan, and BlackSky satellites are then tipped for image collection. 


4. Image Collection: Once tipped, BlackSky satellites collect imagery at the projected coordinates that are most likely to intercept. 


  • BlackSky collects numerous images along the vessel’s projected route to ensure the highest capture probability. This is done to compensate for any variations in vessel speed or changes in heading.


5. Vessel Detection: Once the images are collected and downlinked to the system, the Spectra AI platform begins its analysis. After, a vessel detection model is run to identify and classify any vessels in the images, and detections are cross-referenced against Spire’s AIS data - ultimately determining whether the target vessel is present in the imaging. 


6. Updated Target Geolocation: If the target vessel is, in fact, present and captured in the imagery, Spectra AI updates the vessel position and trajectory based on the new information - ensuring accurate location once a forecast begins. 


The cycle repeats once all six steps are completed, ensuring vessel custody throughout a voyage. If a vessel’s AIS positioning does not match the collected imagery from the process, it most often indicates that the vessel is spoofing its location (falsely altering its AIS positioning). 

Case Study Results & Improvements


This section outlines the initial testing of the MCS prototype, in which it targeted vessels for geolocation efforts using the system. This section will also cover what improvements can be applied to the MCS to address the above limitations to improve the MCS’s ability to accurately and affordably identify, track, and monitor ships of interest.  

Results Overview 


Upon testing the MCS, it successfully captured the targeted ships while docked near seaports and transiting through open water. Targeted ships were captured in around 25% of the image collection executions.


The MCS prototype used a dead-reckoning method (as described above) to forecast vessel positioning, average latency for imaging execution, and single-pass satellite image collection. Considering the approach's limitations (latency, single-pass, etc.), the MCS provided a strong use-case foundation in which further algorithm refinement can be made to improve the service. 

Improvements Overview


The following improvements were identified to enhance the capabilities of the MCS system:


  • Transition from dead-reckoning vessel trajectory to a more advanced route prediction model to ensure images are collected in the correct location.

  • Increase the priority level of MCS imaging to improve latency time between imaging requests and execution.

  • Move from a single-frame to a multi-frame image capture to increase the frame (area) of capture. 


The above improvements would increase the likelihood of capturing a target vessel within a frame of interest and provide better accuracy in vessel monitoring. Each improvement to the MCS is outlined in greater detail in the following sections. 

Improved Route Prediction


As of now, the MCS algorithm uses a dead-reckoning approach to predict vessel positioning and movements. The forecast identifies imaging opportunities for successful geolocation based on where the vessel is expected to transit. By improving the MCS algorithm, the likelihood of a successful interception would increase considerably and provide better overall tip and cue strategy performance.


Dead reckoning uses a vessel’s current positioning and its most recent speed and heading to predict future movements. An alternative to dead reckoning is BlackSky’s heteroscedastic probabilistic routing model, which predicts future positioning as probability densities with confidence intervals that vary over a course. The higher the confidence interval, the more likely a ship will move to the expected position - so estimations can be made with more clarity regarding a voyage. 


These types of advanced models also incorporate information like the most common shipping routes and historical congestion throughout a course - both of which can create varying vessel forecasts. 

Improved Imaging Latency


Another critical component in achieving successful tracking and geolocation involves the time between when a satellite is tipped to execute imaging and when the imaging is collected (latency). By improving (lowering) latency times, location errors are minimized, and route prediction is improved. 


BlackSky uses a planning algorithm to prioritize image collection based on feasibility and incorporates weather conditions, solar illumination, and other factors to prioritize capture. 


The MCS results from this study were executed at a low-priority tier so as not to supersede customer tiers. Consequently, the planning algorithm did not schedule imaging tasks with sufficient lead times - resulting in missed opportunities that could have been avoided. Since the accuracy of vessel forecasts decreased with time, the service collected imaging in the wrong location for vessels that had deviated from the initial predictions. 


By increasing the priority of the image collection to higher tiers, latency times are expected to improve enough to capture images with fewer missed opportunities. 

Improved Area Collection


The MCS captured a single 4km x 6km image from a single satellite on each pass, limiting the study's area of interest. However, the BlackSky constellation can collect numerous images on a single pass (along-track and cross-track), which can be stitched together to improve the ‘area collect’ - which would likely enhance the area collect by up to three times the image collection used. 


To further improvements in imaging collection, BlackSky can also create mission plans to collect images along the forecasted vessel track - which, again, would improve alongside updated route prediction. 

Summary & Conclusion


While this case study proved successful in achieving its targets, the above modifications to the MCS can enhance the detection and geolocation capabilities of the system. With these improvements, the MCS is expected to become a powerful tool for users to improve maritime domain awareness and better strategize solutions to the threats imposed on the maritime domain.

Trusted Data Providers with Tip and Cue Capabilities

Spire Global Space Logo
Iceye Space Services Logo
Kleos Space Logo
Planet Federal Space Logo
bottom of page