Data scientists in disguise
Harriet McQuade, Principal Geospatial Consultant, Atkins
IN a world where we’re using, creating, and consuming more data than ever before, organisations must have a way to understand that data and transform it into actionable intelligence. Data science has become a cornerstone of engineering, yet in the push to strengthen skillsets in AI, machine learning and automation, the geospatial community is also primed to play an increasingly important role.
According to experts’ predictions, total data produced in the world by the end of 2021 had reached 74 zettabytes – a zettabyte being a trillion gigabytes – a colossal number containing 21 zeros. So, it’s clear why we need data scientists to interrogate all this data and extract its inherent value, gaining new insights that can help drive better decision-making. The fourth industrial revolution has most certainly arrived.
Geospatial professionals have been unlocking insights using our own evolving methods of data analysis for more than 25 years. We’ve been helping clients collect, analyse, manage and share their data, putting it to work so that it makes a critical contribution to project safety and performance. Geospatial processes, and the value they unlock, are now embedded within almost every infrastructure project around the world.
GIS is particularly helpful because it lets us simultaneously analyse multiple elements of a problem by using location as a unifying factor between diverse subjects. This enables a more meaningful picture to emerge, supporting rapid, multi-factor, decision-making. Because information is enriched it adds confidence to investment decisions, it helps to boost project efficiency and accuracy, and saves time and money. With the ability to create a holistic view, locational data has an increasingly important role to play in the wider data science landscape.
How geospatial adds to social value
Geospatial processes are already a key tool in the data science ecosystem. One area where it is helping is in adding to social value – because analysing data by location can help to unpick more complex socio-economic problems, making it a useful tool in the government’s ‘levelling-up’ agenda. On recent projects, our geospatial team has combined more traditional spatial analysis of biodiversity, noise and air, with running and cycling tracking data from Strava to identify opportunities to provide more and better green spaces in cities to help improve residents’ mental and physical wellbeing.
Geospatial professionals have been unlocking insights using our own evolving methods of data analysis for more than 25 years.
Where analysing these metrics non-spatially could have resulted in individual implications for each factor, by using location as a unifier spatial data science has revealed how specific interventions could improve multiple metrics simultaneously. Layering these results to view current land use context has produced immediate and powerful visualisations that local councils can use to drill-down in specific areas to aid their decision-making about improving public spaces.
Supporting smoother operations
Spatial analysis is also contributing to data science processes in more complex 3D applications, with Atkins recently supporting a railway network operator client to embrace the power of AI to help smooth its operations. A major task in operating the UK’s railways is the issue of vegetation encroachment; trees and hedges covering or blocking railway lines.
We worked with the client to develop a useful proof-of-concept tool, powered by artificial intelligence, and using LiDAR technology plus hyperspectral imaging, that can highlight features that are visible and invisible to the human eye. Together, these tools were used to automatically identify problematic areas where branches are straying into the complex 3D rail zone.
Locational data has long been used to track and monitor the issue but combining it into a model powered by artificial intelligence created a tool that can predict and detect any vegetation that could cause a problem. Using feature manipulation engine (FME) software to set out the different zones, extending them where there were overhead lines or signal equipment, and increasing height detection proportionate to width, the resultant data science was able to detect any vegetation that could potentially fall from greater heights. Finally, the resulting data was fed into our digital twin platform, CIRRUSInsite, that allows users to view and analyse big data instantly.
A shift up in gear for predictive maintenance
Armed with this information, the rail network operator’s decision makers now have access to data that helps them proactively plan clearance work, reducing cost and accidental disruption to the network, and improving safety for trackside staff who now need to go on fewer site visits. Our methodology for this issue to the rail sector serves as a good blueprint for other sectors – including for built infrastructure assets.
It represents a shift up in gear for how we can perform risk management and predictive maintenance. It is a great example of what geospatial data science can achieve when it is blended with other data science disciplines.
Strong collaboration, such as bringing together data science disciplines, areas and expertise – we call this our data intelligence division – is key to ensuring locational data science achieves its full potential and helps decision-makers opt for the best choices. We can now fix problems that we never dreamed of fixing through the smart manipulation of new data. In my view, geospatial is a key data science that forms a critical part of the puzzle to turn this data into intelligence.
Harriet McQuade, Principal Geospatial Consultant, Atkins
Sources
https:// financesonline.com/how-much-data-is-created-every-day/ analysis by IDC & Statista, 2020
https:// www.sciencedirect.com/topics/earth-and-planetary-sciences/hyperspectral-imaging