Star search program used to trail whale sharks

Scientists have used a technique for pinpointing an object's location in the night sky to identify and track whale sharks in the Earth's oceans.

Current whale shark tracking methods include passive tagging of the sharks with serial numbers – which can fall off or irritate the sharks – and tags that can be tracked by satellite, which fail below a certain depth and have a limited battery life.

So marine biologists, like Brad Norman at the Centre for Fish and Fisheries Research at Murdoch University, Australia, have ended up using each whale's unique pattern of spots to visually identify individuals from photos – and ultimately track their movements over several years. But the process is laborious: Norman often resorted to spreading all the pictures on the floor to find similarities in the spot patterns.

Then optical astronomer Gijs Nelemans, of Radboud University, the Netherlands, introduced them to the Groth algorithm, used to compare triangular patterns among stars in a new photo to patterns in an existing image. If a possible new bright spot was seen in the sky – a supernova, for example – the program would give its precise location in the heavens.

Norman and colleagues – software specialist Jason Holmberg, based in Seattle, Washington, and astrophysicist Zaven Arzoumanian of the Universities Space Research Association, located at NASA’s Goddard Space Flight Centre in Greenbelt, Maryland, both in the US – took a version of this program and replaced the stars with whale shark spots. They fed it with over 1500 photos from the web-based ECOCEAN Whale Shark Photo-identification Library, which contains pictures from marine biologists, and others submitted by diving enthusiasts.

In a test using known matches, the new program correctly matched shark photos with the same spot pattern over 90% of the time. The program even made matches that Norman had initially missed.

The program has since made more than 100 new matches that were later verified manually. "What we've done is eliminate the need to scatter all the images on the floor," Arzoumanian told New Scientist.

So far, the team has not been able to account for the effect of compressed or stretched spot patterns if the fish is swimming at an angle to the camera. The researchers are now working with 3D-surface-modelling to make the system even better at identifying the fish.

Tracking the whale shark's movements over many years could tell biologists more about the species' population trends and migration routes and whether wildlife refuges are having the intended protective effect on whale shark populations.

The researchers hope that the method could be applied to tracking other creatures. Lions, for example, have small dark spots where their whiskers attach to their noses, which could also be used for tracking them.

Source: www.newscientist.com