A team of Kiwi scientists is using pictures taken by satellites in orbit to enable farmers to spot pollution on their land.
New Zealand scientists are looking to put satellite imagery to good use by identifying pollution on Kiwi farms, helping make tomorrow a tiny bit greener.
University of Auckland’s Intelligent Vision System lab (IVS) is developing a way to spot pollution by applying computer vision techniques to photos available on aerospace data sets. Computer vision, the lab’s speciality, is about using algorithms to extract information from images. With satellites completing an orbit around the Earth as frequently as every 1.5 hours, the photos they take can be compared to identify developing issues.
Research fellow Trevor Gee, the project’s technical lead, said the goal is to help farmers identify pollution that they’re unaware of, warning them when things start to go south. “We can detect from changes [in the images] that something sinister is happening – the river is starting to go green or there’s a build-up of strange objects in a particular location – and that can raise an alarm for investigation.”
Not all changes are going to be about toxic sludge in waterways. Many will be easily dismissed: sea waves in a different place from yesterday, or a rainy week creating water deposits around the land. To separate this statistical “noise” from the areas of concern, the team is building a classification system that labels differences as either dangerous or benign, said Gee.
“The typical way to construct a classifier is to build a statistical model from lots of training example images – a technique known as machine learning, which can be seen as a type of artificial intelligence.”
Having this early warning system could mean scientists will be able to identify and mitigate problems fast, rather than having pollution build up and only come to light when it’s an issue, said Gee. “What tends to happen is that pollution issues tend to build up over time and no one’s really aware of them.”
to our journalism!Find Out More
Computer vision is a development stemming from the field of photogrammetry, the use of photography in creating maps. Gee says the key difference between them is that computer vision is more accessible. “Photogrammetry was a heavily structured field, so if you wanted to make a map on the ground you had to plan a flight path, you had to design exactly which cameras you were going to use [and] all the parameters,” said Gee. With their project, farmers will simply access the information they need through a mobile app.
With water pollution being one of the biggest concerns of New Zealanders, The Spinoff asked IVS if satellites could help detect river pollution before it got too dire. The answer is probably not. In rivers, the pollution may be too diluted for the satellites to detect, especially if the photos are few and far in between. “In short, for rivers, a fast response is paramount and satellite imagery detection may only work if we can ensure several flyovers of any given satellites on a daily basis,” said Patrice Delmas, lab director at IVS.
But turning to the stars is just one option in identifying pollutant sources. The second stage of IVS’s project uses equipment a little closer to home. Delmas said the lab was looking to create a mobile-based alert system where users send a pollution warning, after which drones are employed to fly over and photograph the area in question.
The IVS lab is one of the 18 teams groups taking part in this year’s Aerospace Challenge, a competition to help agriculture and sustainability in New Zealand. The winning team will get a cash prize of $30,000 and support from industry to help commercialise their idea. This year’s task is to develop a way of using satellites or other unmanned aircraft technology to detect or assess pollution in water or soil.
The Spinoff’s science content is made possible thanks to the support of The MacDiarmid Institute for Advanced Materials and Nanotechnology, a national institute devoted to scientific research.
The Spinoff Daily gets you all the days' best reading in one handy package, fresh to your inbox Monday-Friday at 5pm.