Digital twins for kākāpō and AI to track kiwi calls: How tech is helping save NZ’s endangered birds
- 22 August, 2019 06:00
The kākāpō is a large, flightless, nocturnal, ground-dwelling parrot endemic to New Zealand
My vision is the IP we generate in developing these sorts of services is shared with conservation industries all over the world
There are 200 kākāpō in New Zealand, and each and every one of them is able to be monitored by the Department of Conservation.
He says the ICT team at DoC worked closely with the field staff to determine how best to support the conservation of the kākāpō, one of the rarest parrots in the world, and another endangered bird species, the kiwi.
“I have worked on conservation all of my life,” shares Edginton. “Now, I tell people I do conservation with different tools.”
According to DOC, every kākāpō today wears a smart transmitter that sends out a radio signal. The signal describes the bird’s position, whether it’s still alive, and how much battery the transmitter has left.
Males and females wear different versions of the transmitter, which add behavioural information to the signal.
For the kākāpō project, he discloses that DoC had a 20-year-old database that was “definitely end of life”.
With the breeding programme coming out, the team was concerned the database would not be able to make it through.
It is a privilege for me and my team to work on these projects
So, they built a platform that works with other versions of the database that could share data in the field. The data is secure and always backed up.
“We worked really closely with the business. We were out in the field with them and delivered what I think is a good system for them.”
The new database was completed by his team from the beginning of 2018 up to September the same year.
Edginton says data management has to be made simple for people in the field, and for them to be able to share the data more widely across other related services.
There is a mesh network out in the forest that is connecting the weighing stations and feeding stations, and sending the information back to the database.
“When the tagged bird stands on a scale, we know how heavy it is, and also how much food to give to it,” says Edginton.
“That has been really important, because we have a very good profile about the breeding capabilities of the kākāpō, especially of the females out there.”
“We have a day-by-day view of how each of the bird is doing whenever they turn up into one of the weighing stations and feeding stations,” he adds.
Part of the wider kākāpō project is making an egg with sensors in it that informs the rangers about the conditions of the other eggs in the nest. “Are they moving? What temperature are they at?
“We can really have a really close view of eggs when they are in the nest or in the incubation box,” says Edginton.
In the case of the kiwi, there are around 68,000 of these flightless birds left and they continue to be threatened by predators, such as stoats and dogs.
Edginton says over the last 20 years, there has been a lot of audio recording of kiwis out in the field.
“We literally have terabytes and terabytes of this data. These have not been decoded,” he explains.
DoC asked analytics firm Qrious to decode this data using AI and machine learning, and develop for them a way to count kiwi calls.
They have been able to teach the machine to tell the difference among the sounds on a file and to ensure it is a kiwi call.
“It is going to be smarter and smarter the more data we put through it,” he states.
He says the project shows the power of using the core AI and machine learning services of AWS.
“What I want to do with this is to create a view of the soundscape of the environment of New Zealand.”
“I want this code and the data to be open source, so any conservation agency around the world can do the same and start building their own soundscape based on the species they want to count, and to tell them something about the condition of that environment.”
The CIO as social entrepreneur
Edginton says the most important thing is that the DoC teams not only contribute to ensure the species survives, but also share the IP of the project - data and the code - on open source.
The next step is to publish the database on GitHub.
“For other countries that are struggling with rare and threatened birds, they can use this database and this design and all the data to manage their own threatened species,” Edginton explains.
For instance, the code from the kiwi call system is now being used for other purposes across the ditch, such as tracking koala recordings to track and manage koala populations near Brisbane.
He says their work has applications for other business models, too.
“I would say a kākāpō is an asset and we are managing that asset, and the data that is sitting there is also an asset we can leverage. We can model from that data how successful we are and others can use that model, as well.”
“The lessons from the programme are probably directly applicable to the agriculture industry, in managing rare species of rice, for example,” he adds. “How are they looked after? What can we do to protect the food sources around the world?”
His message to CIO colleagues?
“I am interested in sharing this as a social enterprise type of story,” he states.
“Conservation affects us all in every part of our community and realising how we can share both our IP and data to build a community of interest and practice that will actually help save the 2000 seriously threatened species that we have in New Zealand.”
“It is a privilege for me and my team to work on these projects, and to guarantee the succession of these critically endangered species.”
Sign up for CIO newsletters for regular updates on CIO news, career tips, views and events. Follow CIO New Zealand on Twitter:@cio_nz Join us on Facebook.