Post Catastrophe Imagery and AI-derived property damage and condition data unite to help insurers process customer claims more efficiently.
In October 2023, Sydney hosted the tech, culture, innovation and music festival that has been an feature of life in Austin, Texas for 37 years: South by Southwest, SXSW – or ‘South by’ as its known by devotees. SXSW Sydney brought together thinkers and innovators to share, learn and collaborate. At the Tech & Innovation expo, the Discovery stage hosted science communicators like Dr Karl Kruszelnicki, Adam Spencer, astrophysicist Karlie Noon, nanotechnologist and tech innovator, Dr Michelle Dickinson. It also hosted two inspiring Nearmappers: Dr Michael Bewley, Vice President, AI & Computer Vision, and Dr Kitty Lo, Principal Data Scientist.
Building on the key SXSW Sydney theme of ‘Discovery’, Mike and Kitty took to the Discovery Stage with their presentation, Mapping our Future with AI, taking the audience on a journey through the era of intelligent mapping and geospatial analysis. With multiple sessions across the week featuring versions of Generative AI, Mike Bewley introduced his talk by letting the audience know that he wouldn’t be talking about the Chat-GPT side of AI, as “there’s so much more going on.” Watch the presentation
The presentation began with Mike sharing his 25-year career progression, including his admiration for Principal Component Analysis, PCA, which played a role in helping him create analytics to detect Alzheimers in brain scans, and to develop technology that helped people with hearing impairment receive signals to the brain through Cochlear implants.
Mike wowed the audience with a visually stunning representations of Nearmap AI in action, highlighting the uses that the technology enables, at scale, such as identifying buildings, construction sites, construction cranes, lawns, cars, swimming pools, solar panels, trees, power poles, light poles, trees overhanging buildings. “We can do this several times on a property every year and do it on the scale of a hundred million properties,” said Mike.
Dr Kitty Lo continued the story by demonstrating how Nearmap AI is being used for disaster preparation and recovery, with imagery and AI data layers providing insights across affected areas, often within days of being surveyed after a catastrophe.
Kitty explained how AI can help define the risk factors which make buildings or locations more vulnerable to natural disasters, and how AI can automatically identify damaged buildings, classifying the level of damage at scale, from a single building to entire neighbourhoods and regions. Kitty demonstrated how the technology identified buildings destroyed or damaged in the Australian 2019 south coast bushfires, and debris on the ground prior to the 2022 Mill Fire in California.
AI data layers can provide insights over time, across large areas, with Nearmap covering up to 95% of the Australian population. An example Kitty shared was when the Nearmap team asked the question: over the last 15 years, which area in Australia had the highest number of tarpaulins?
Analysing the Nearmap AI data uncovered Berowra, NSW as the area with the most tarpaulins at any one time – due to a large hailstorm that struck the area in 2018. A series of images depicted tarpaulins highlighted by AI – and showed the gradual decrease in the number of roofs under tarpaulins in the months and years after the storm.
Dr Mike Bewley returned to the stage with glimpse at just some of the work Nearmap has been doing using AI data to assess tree canopy cover – a hot topic in our increasingly hotter suburbs. Looking at the national data on 5,000 suburbs across Australia, Mike showed the nation on a page, in the form of a bubble graph showing cities, by colour, and how much of the area is covered by trees compared to the percentage covered by buildings, comparing data for 2019 and 2023.
Applying the same data in the U.S. (from surveys capturing around 110 million buildings) showed that there are more suburbs and towns in the U.S. with a higher percentage of tree canopy coverage than Australia. Mike explained that this is due to larger average property lot sizes in the United States compared to Australia, where new residences tend to build right to the boundaries, with very little room for green space or large, mature trees.
However, one encouraging finding from the Nearmap residential tree canopy study is that although new, in-fill suburbs get bad press at the outset for their lack of tree cover, often they’re built onto disused farmland, which didn’t contain many tress to begin with. As new suburbs are created, tree cover remains low for several years, as saplings are planted. But over time, the Nearmap analysis is uncovering some positive indications: given the right conditions, tree canopy increases over time, as the trees mature and take hold in the environment. For many of these new suburbs, this growing tree canopy will create more shade, and slightly reduce overall temperature in the surrounding areas. Trees also impact mental health, physical health, property prices, fire risk, urban heat islands and a range of other factors.
Mike shared another positive angle that Nearmap AI has uncovered: that most newly built suburbs show an almost 100% uptake of solar panels on all residential and most commercial buildings. As a general rule, new suburb builds will often include solar as included, by default. This outranks older, more established suburbs with property owners might be more concerned about elements including aesthetics, cost and practicality, when it comes to considering whether to retrofit solar onto a home.
Dr Mike Bewley and Dr Kitty Lo’s session asked the questions that will help us create more sustainable neighbourhoods.
What kind of cities do we want to live in? What should those cities look like?
How are things changing, and how do we want them to change?