ASTRIIS to detect and classify oil spills and other ocean hazards

With a large variety of resources coming from the ocean, there is a growing need for services of surveillance and monitoring the surface of seawater. Hazards such as oil spills, toxic algae and plastic patches constitute a looming threat towards the ocean’s health, disturbing its environment and biodiversity, while also impacting blue economy industries like fisheries and aquaculture. Having means to detect, classify, and monitor such hazards, will not only help reduce their damages, but may also help to prevent them and to identify who is responsible.

ASTRIIS is developing a solution consisting of a merge of different sensors and platforms. Under the lead of Tekever, such activity is focused on sensors beyond the standard RGB cameras, namely SAR (synthetic aperture radar) and a hyperspectral camera.

SAR is a type of sensor meant to be operated airborne and serves to obtain aerial imagery either from an aircraft or a satellite. This technology makes use of radio waves, some signal processing, and the movement of the airborne vehicle itself to acquire a detailed relief mapping of the Earth’s surface. The method is fast and can cover large areas quickly.

In a marine context, the texture of the surface of the seawater can reveal the presence of contaminants. For example, since oil has a higher viscosity than water, in an oil spill scenario the wind will not create wavelets on the oil. This way, in SAR, a spill will appear as a smoother area on an otherwise rougher ocean surface.

A SAR image of an oil spill. Lighter areas represent changes in relief, the oil spill is the darker smoother area. (From: ESA Ocean Virtual Laboratory).
A SAR image of an oil spill. Lighter areas represent changes in relief, the oil spill is the darker smoother area. (From: ESA Ocean Virtual Laboratory).

 

Other hazards, such as harmful algae blooms (HAB), plastics or even animal feces can also cause anomalies in the surface of the ocean. This way, using SAR alone it would not be possible to tell these apart. To make this distinction, ASTRIIS proposes using an auxiliary method: hyperspectral imaging.

Standard RGB cameras, such as digital photography cameras acquire three values of data per pixel: red, green, and blue. However, the real light spectrum is continuous, with the colours existing in-between and as a combination of these. A multispectral or hyperspectral camera subdivides the light spectrum and acquires from tens to hundreds of values per pixel. This way, there is the potential to distinguish between materials even if these appear as the same colour to the human eye. This process nevertheless produces large quantities of data, which is harder to transmit and process if larger areas are covered.

 

An AR5 system with a SAR sensor equipped. The antennae for the sensor are embedded in the dark rectangles on the sides of the fuselage. (From: Tekever)
An AR5 system with a SAR sensor equipped. The antennae for the sensor are embedded in the dark rectangles on the sides of the fuselage. (From: Tekever)

 

This way, ASTRIIS proposes a combination of techniques, where a larger area is scanned using SAR, and once hazards are detected, using hyperspectral imaging solely on the area of interest, the event is classified. The sensors are to be mounted to an AR5 system; an unmanned air vehicle developed by Tekever with an autonomy of 12 hours.

 

An AR5 system with a SAR sensor equipped. The antennae for the sensor are embedded in the dark rectangles on the sides of the fuselage. (From: Tekever)
An AR5 system with a SAR sensor equipped. The antennae for the sensor are embedded in the dark rectangles on the sides of the fuselage. (From: Tekever)

Stay tuned to the ASTRIIS website and LinkedIn for further developments on this and other ongoing project activities.