The world is warming and nature is in crisis - but the pathway to Net Zero and recovery runs through digital, claims UK-headquartered building design specialist Arup.
Core to providing solutions to the climate crisis, says the consultancy, is a synthesis of cloud, analytics, and machine learning.
These tools - as delivered through its ongoing partnership with Amazon Web Services - are at the heart of two new Green Arup projects that involve ‘UHI’ (urban heat island) mapping and habitat analysis.
These problem areas and others are being targeted by a new Arup Data Platform, powered by a number of Amazon Web Services (AWS) resources and applications.
The context for the UHI application is that cities are starting to experience warmer temperatures than their rural surroundings.
If city managers could see exactly where in their cities heat is spiking most, they could intervene by planting trees or even painting streets white to increase albedo.
An opportunity to possibly work on a block-by-block basis - or, in the world of geospatial engineering, at the ‘hexagon’ map level - could mean remedial work could be even more effective, says the company’s Digital Services Lead at Arup, Will Cavendish.
This capability is now being offered to Arup clients via a combination of ‘raster’ satellite imagery, Earth observation, as well as map (‘vector’) and remote-sensing data to closely model and quantify the factors that influence UHIs.
Using a mixture of data inputs to look at urban heat islands isn't new, says Cavendish’s colleague, Damien McCloud, Director and Business Lead for Geospatial and Earth Observation.
But, he adds:
The difference between ten years ago and now that back then we used one layer of data land surface temperature. What we are delivering here brings together vast quantities and lots of different types of data to e able to understand the space better.
Adding in new data sources at much greater detail includes street surface reflectivity, building emissions and heights, shadows, and the specific heat difference between tree-lined and non-tree-lined streets.
That difference, says McCloud, can be two degrees in a heat event - so having multiple ways to analyze UHIs becomes a much more powerful tool.
Having all of these datasets brought together is really key to providing insights that master planners can use and understand and to model what different interventions do - how they can work together to bring that temperature down and mitigate climate change.
Examples of the benefits of having such UHI tools, says the company, include helping local governments quickly identify buildings and populations at risk from extreme heat events. Now, says the company, processing of UHI case data has gone from two weeks to under an hour.
Another new Arup service making use of cloud and machine learning is accurate biodiversity mapping. Again, various layers of data and satellite imagery produce a map of habitat distinctiveness to not just calculate and quantify biodiversity scores for scenarios interventions, but also support long-term monitoring of outcomes.
That matters, as more and more countries now ask for decades of environmental impact analysis of a new building or infrastructure project and, as Arup people like to joke, there are only so many graduates in the world you can get to walk around sites to do that.
As McCloud notes:
At the heart of everything sustainable and development is measuring and recording; we’ve got to understand where we are, understand what difference interventions make. With nature credits and trading in particular, it's about validation, assessment and monitoring of those credits to give us better understanding to allow us to make better decisions.
Using Earth observation data means you can monitor what's happening and what interventions are and are not working so you can start to build a body of evidence that allows you to move forward that's based on actuals, not anecdotes - and at scale.
The move to geospatial thinking
Both Arup’s UHI and habitat services are examples of geospatial engineering.
This is a way to move beyond even the most sophisticated Geographical Information System by adding in multiple other data sources.
In Arup’s case, geospatial is being delivered by use of both AWS data sources and data tools.
This includes access via AWS to Landsat 8 and Sentinel-2 Earth observation satellite, global open geographic database OpenStreetMap data, as well as other libraries.
But the other half of the Amazon geospatial equation - now productized as a special version of its SageMaker cloud machine-learning platform - is a set of tools and utilities to clean up and organize this data.
A big help, says McCloud, is automatic removal of cloud and shadow pixels from satellite imagery, as well as automatic identification of different land classes and surfacing of hidden ground information or extraction of landmarks.
Other SageMaker Geospatial tools - which Arup was a partner in co-creating - include a number of other geospatial visualization tools and pre-trained models.
This data is then used to create specific Machine Learning models which Arup uses to help clients like the capital of Albania, Tirana - where a new orbital forest will deliver $75m worth of sequestered carbon benefit over 40 years, for example.
What our new geospatial machine learning tool allows us to do is to have all of those models in there and run them quicker than we've ever been able to do before to get to a more robust answer.
Using tech to renew the built-in environment
Cavendish sums up the benefit of cloud, analytics and machine learning geospatial for his company and says:
We're trying to use the best of technology, data and digital to try and solve these problems, like designing lower carbon buildings. We’ve got a 12-storey building in Amsterdam that is just made of wood and steel, but you need really complex algorithms to understand the structural strength of that approach to make it work.
We're using the same approach to prolong the lifespan of assets, because we want to make sure we can use the stuff we've got as long as possible. You need super cool machine learning capabilities to analyze existing buildings to figure out their structural state and make them last longer.
We need a smorgasbord of technologies to solve our carbon and nature issues in the built environment.