Artificial intelligence, virtual reality, and space
satellites are just some of the new technologies being used
to increase Aotearoa New Zealand’s resilience to natural
hazards according to EQC Toka Tū Ake.
“Our
country’s scientists are world leaders in natural hazards
research, thanks in part to our country’s standing as one
of the riskiest places on earth,” says Dr Natalie Balfour,
EQC Head of Research.
This Tech Week, New Zealanders
are celebrating home-grown advances in technology that make
our nation a better place.
“Our researchers and
scientists are doing just that by using cutting-edge
technology to help people understand their natural hazard
risks and better prepare their homes and families for future
events,” says Balfour.
“From harnessing artificial
intelligence to detect earthquakes, to using virtual reality
to create an immersive experience of flood risks, EQC is
proud to fund exciting research that will influence decision
making and resilience for years to come,” she
says.
Space satellites predict future
landslides
Landslides are New Zealand’s
deadliest, and amongst its costliest, natural hazards.
Extreme rainfall in Auckland and Gisborne in 2023 triggered
devastating slips, causing billions in damage and loss of
life.
Scientists are now revisiting the worst affected
sites to investigate whether the land gave warning signs
prior to slipping, and the likelihood of future
events.
“Until these problems are addressed, it’s
difficult for people to return to their homes, rebuild, or
return to any semblance of normal life,” says Associate
Professor Martin Brook from the University of Auckland, who
is leading this work with support from an EQC Biennial
grant.
Advertisement – scroll to continue reading
Brook is using InSAR technology, whereby
space-bound satellites map the earth’s surface to detect
ground movement on a millimetre-scale.
It’s not the
usual way to study landslides in New Zealand, which
typically involves physically placing ground motion sensors
on affected land. While effective, it’s a difficult and
dangerous technique.
“InSAR can give us a much
better understanding of landslide hazards at a street or
even individual land parcel level, without stepping foot on
the ground.”
Brook will analyse data beamed from the
Sentenil-1 satellite, launched by the European Space Agency
in 2014 as part of its Copernicus Programme.
By
comparing detailed satellite images taken over the same
geographic areas, before and directly after the storms,
Brook hopes to glean insights into what caused the ground to
slip and whether the area is more vulnerable to slips in the
future.
This is one of the first uses of InSAR in New
Zealand, but Brook explains the technology is already
prevalent overseas, where it regularly provides early
warning of ongoing slips.
“The most amazing and
extensive example is the European Ground Motion Service
website portal, which allows you to zoom in anywhere in
Europe or the British Isles and see ground motion for the
last few years.”
Brook hopes his project encourages
something similar to be implemented in New Zealand.
Landslides are such a widespread hazard that many people
stand to benefit from early warning systems including
homeowners, business, planners and insurers.
“A
better understanding of landslides can help us constrain the
hazard and define risk more accurately.”
Brook sums
up his approach as, “titiro whakamuri kia anga whakamua
– looking at the past, using InSAR, to move
forward.”
AI paints a clearer picture of
New Zealand’s seismic activity
Although
it may not feel like it, the ground beneath our feet is in
constant motion. Sensors picking up ground vibrations
generate reems of data, which are trawled through by
seismologists searching for patterns that indicate an
earthquake might have occurred.
Big earthquakes are
easy to detect, but small ‘microearthquakes’ can be hard
to pick out from vibrations caused by noise from things like
passing trucks, crashing ocean waves, or the wind shaking
trees. As a result, many microearthquakes go
undetected.
Dr Calum Chamberlain, a Senior Lecturer in
Geophysics at Victoria University of Wellington, explains
why this is a problem: “Microearthquakes are an incredibly
useful natural monitor for what is going on inside the
Earth.”
Earthquakes occur when the ground is subjected
to so much force that it fractures or breaks. Small
earthquakes happen along the same faults, and as a result of
the same stresses (such as compression squeezing rocks
together, or tension pulling them apart), as big
earthquakes.
“To know which faults are likely to fail,
we need to know where the faults are and the direction of
the stresses acting within the earth.”
Since AI is
famously good at detecting known patterns from large
datasets, seismologists like Chamberlain have started using
it for earthquake detection.
With funding support from
EQC, Chamberlain’s group is applying AI to revisit
historic earthquake catalogues in New Zealand and find
evidence of previously undetected
microquakes.
“We’re detecting many earthquakes
that were missed within the decades of data collected by
GeoNet, and as a result painting a more complete picture of
where and how frequently earthquakes occur across New
Zealand. This will give us a better idea of where faults are
and how they interact with each other, helping inform us on
how future large earthquakes might work.”
Over the
years, Victoria University’s school of Geography,
Environment and Earth Sciences, where Chamberlain works, has
developed exceptionally high-quality earthquake catalogues
against which researchers can use as a benchmark against
which to validate the new AI tools.
“Our biggest
contribution globally in this space has been identifying
issues with applying the AI blindly and developing
workarounds which make the tools more robust and
reliable.”
As a result of his and others’ efforts,
Chamberlain expects AI to get better and better at
earthquake detection.
“We’re almost at the point
where we won’t need any human intervention to detect
earthquakes. However, detection is only a small part of what
we do as geophysicists. Hopefully, by making more use of AI
in basic detection tasks, human intelligence can be freed up
to think more about the physics behind the
earthquakes.”
Virtual reality brings
flood risk to life
If VR could bring the
risk of flooding to life, would it spur action by
communities to prepare for, and even prevent, future
events?
That’s the question being explored by
University of Canterbury’s Professor Matthew Wilson and
leaders from Ngāi Tahu, with funding support from an EQC
Biennial grant.
What we know about flood risk is
currently captured in detailed 2D flood models, which offer
a treasure trove of information about where floods are
likely to occur and how, but they’re hard for most people
to decipher.
Wilson says, “What excites me about VR
is its ability to break down barriers of communication by
providing an immersive understanding of
risk.”
Wilson will turn existing flood models of key
South Island sites into interactive VR flood simulations,
and test whether it’s a useful way to enhance people’s
understanding of flood risk.
It will be the first
open-source, accessible software of its kind in New Zealand.
“Several international companies have built VR flood
models, but high license fees keep them out of public
reach.”
Wilson is currently working closely with
leaders from Ngāi Tahu to identify appropriate sites within
their rohe to virtualise.
The new VR models will be
tested by Ngāi Tahu whanau, who will provide feedback on
whether the experience aided their understanding of
potential severity and consequences of future
floods.
If the tech proves useful, Wilson hopes to
extend the VR map to more parts of the country and even
introduce AI to generate dynamic scenarios that respond to
user input.
“My ambition is for people to be able to
dynamically make changes to their environment in the VR
world, and see how it impacts future flood risk,” says
Wilson.
Iain Gover from Te Rūnanga o Ngāi Tahu says
“it’s fantastic to have Ngāi Tahu involved in driving
this research, with technical firepower provided by Matt and
the University of Canterbury. “
Gover is especially
interested in how nature-based mitigation strategies, such
as wetland restoration or native tree planting, can mitigate
flood risk.
“It may lead to a double win, mitigating
flood risk and improving ecosystems at the same time.
Hopefully, VR will allow us to see more clearly how changes
to landcover and land use change flood risk” adds
Gover.
Advertisement – scroll to continue reading
