Censusing Animal Populations from Space

Antarctic ice seals are the most abundant mammals in the Southern Ocean, but surprising little is known about their populations or distribution. The ice seals are defined as those that breed on sea-ice and there are four species in the Antarctic; Weddell, crabeater, Ross and leopard seals. Because of the icy, dynamic and transient ice on which they breed, studying these species has always been a challenge and current knowledge of many aspects of their lives, including population trajectories and habitat preferences is sketchy.  But the world which ice seals inhabit is changing. The sea ice, on which they are so dependent, is forecast to rapidly diminish with future climate change. Getting robust population estimates, trajectories and distribution maps of the seals and understanding their habitat preferences is therefore vital if we are to quantify how vulnerable the animals are and what effective protection measures could be put in place to conserve them.

That is where the Censusing Animal Populations from Space (CAPS) project comes in. In the past, planes, helicopters and ships have been used as the platform to count the seals, but these traditional methods of survey are slow and expensive, and rarely cover enough of the shifting sea ice habitat to give us robust estimates of the population. The CAPS project is using a different approach. The main initial focus is to acquire population estimates of pack ice seals using very high resolution satellite imagery. The satellites used acquire images with a pixel size of 31-50 cm on the ground and seals are clearly identifiable as clear dark oblongs on the white ice.


A 30cm resolution WorldView3 satellite image showing Weddell Seals near the South Orkney Islands. The red on the ice highlights a freshly born pup. Image curtsey of Maxar Technologies.

A multinational group of scientist is working on the problem, assessing the ability of the satellites to count the different seal species in several Antarctic regions. The seals range over huge areas so we are testing various approaches to counting, included crowd sourcing, machine learning and deep Learning. Our hope is that we can use this new technology to understand seal’s populations and why they prefer certain areas. This will then help us to understand how these animals will cope with the challenges of climate change over the decades to come.

Copyright 2015 Southern Ocean Observing System. All rights reserved.