Using Earth observation data and machine learning to prevent ship-whale collisions

Whales - especially great whales - play an indispensable role in mitigating climate change. On average, a whale accumulates about 33 tonnes of carbon dioxide (CO₂) in its lifespan. Whales also support the growth of phytoplankton, which stores 40% of all CO₂ produced. But these majestic marine mammals that capture carbon from the atmosphere, catalyze oxygen production and maintain the biodiversity of oceans are at constant risk of fatal collisions with ships—the leading cause of whale mortality.

One of the biggest challenges is a lack of data about the location, habits and routes of whales.

As part of AI4Whales, a program supported by the ESA Network of Resources Initiative, CGI in Germany developed a proof-of-concept (POC) that uses machine learning and Earth observation (EO) data to detect and track whales. The POC was developed pro bono at the Innovation Technology Campus in Munich, a co-innovation space where CGI experts collaborate with local students on new technologies and innovative ways to use them.

Based on a sample of aerial images of whales collected by drones, the solution uses machine learning to scan the ocean and, with the help of satellite images, automatically detects whales swimming on the surface. Our cloud-based scalable production platform for EO and geospatial services, CGI GeoData360, tracks individual whales and identifies areas of high concentration and migration routes to propose alternative shipping routes to maritime companies.

The scalable, non-invasive solution can be extended to help study and protect numerous marine and land species by using various data sources, such as drones, weather balloons and acoustic sensors, along with IoT technology and LowRa communication.