Using high-resolution satellite imagery,\u00a0Peter Fretwell\u00a0and his team began monitoring animal populations, creating the world\u2019s first census of species taken from space in 2009. Then went on to repeat it. \u201cWe started with penguins, then have gone on to look at whales, then seals \u2013 we haven\u2019t published much on seals yet. Now, we are working on albatross.\u201d \u2013 said, Fretwell.<\/p>\n
Fretwell worked in the Antarctic for 15 years, initially as a cartographer, mapping the area for pilots. He accidentally found himself one of the world\u2019s leading experts on animal population monitoring from satellite data while trying to locate bird colonies in Antarctica with greater accuracy. \u201cPilots need to know where the runways are, where the mountains are, and where the bird colonies are in order to avoid them.\u201d His team found the emperor penguins\u00a0were poorly geo-referenced, as they breed in extremely remote places and move around a lot.<\/p>\n
\u201cI was looking for one specific colony that I thought was near Halley station, but our knowledge wasn\u2019t good enough for the mapping necessary. While remapping the coastline, I remember noticing a small brown stain on the sea line in the Landsat imagery, which turned out to be the guano from the penguin colony.\u201d<\/p>\n
From there, using open satellite data, Fretwell and his team at the British Antarctic Survey were able to locate previous unknown colonies. In 2012 using the same methodology, the team mapped penguin huddles \u2013 which can contains hundreds of birds \u2013\u00a0 then looked to obtain higher resolution satellite imagery for the areas where colonies were located. The remotely-sensed images were then analysed using a supervised classification method to separate penguins from snow, shadows, and guano and actual counts of penguins from eleven ground truthing sites were used to convert these classified areas into numbers of penguins using a robust regression algorithm. Peter Fretwell and colleague Michelle LaRue estimated the population size to be around 595,000 penguins \u2013\u00a0twice as much as previously thought.<\/p>\n
Continuing his work, Fretwell moved on to whale counting. In 2014, Fretwell and his team came up with an algorithm to automatically identify \u2019whale-like\u2019 features in the ocean. The early version of the automated approach found 89% of probable whales identified in the manual count.<\/p>\n
Many whale species are currently endangered. Knowing more about their behaviours and population size would help protect them. However, previous counts were conducted from a shore position, from the deck of a ship, or from a plane, rendering them necessarily narrow in scope. Additionally, for many species breeding grounds are often within the 200-nautical mile radius of coastal waters, making it difficult to track, in part due to political or military restrictions.<\/p>\n
With the help of the International Whale Commission, the team decided to focus on 4 whale species: fin whales, southern right whales, humpback whales, and grey whales and analyze how difficult each specie would be to map.<\/p>\n
From wildlife conservation to environmental monitoring, satellite imagery is opening new possibilities for understanding our planet at an unprecedented scale. Through our Forward Deployed Engineer (FDE) approach, we help organizations identify the right Earth Observation technologies and develop practical solutions that turn satellite data into actionable insights.<\/p>\n