Skip to content
Newsroom team

Geospatial AI - unlocking insight from data

With artificial intelligence likely to be the most impactful technology driver of change in the geospatial industry, what should companies and organisations be thinking about to compete?

At Ordnance Survey we’ve learnt that AI isn’t necessarily a silver bullet, but instead a tool for improving operations to increase process efficiencies and extract more meaningful insights from data. 

In a project to track urbanisation in Zambia, we automatically extract features such as buildings, roads and water from imagery data via machine learning. The algorithm learns the characteristics of objects by being trained on human-labelled data. This significantly reduces the time required to create a map because it decreases the amount of manual processing. To help train the algorithm, we used the vast and rich data set of Britain and its features.

"Its now possible for AI processes to create a map of any area quickly and accurately from satellite imagery."

The use of Earth Observation data such as satellite imagery is a growing area for the application of ML / AI techniques. Partly because of the falling costs associated with capturing high-definition imagery, lidar and other map data inputs. But also the proliferation of data combined with advances in algorithm development and high temporal resolution are enabling sophisticated use cases delivering precision, accuracy and timely insights.

One use case developed by the rapid prototyping team at OS, involves monitoring peatland, an important carbon store in the UK. By combining multiple layers of satellite imagery with data on temperature, vegetation and albedo, its possible to cost-effectively monitor changes to peatland health. 

 

Meandering river through wetlands
Any business focussing on land management, or important environment areas such as land use, habitat loss, vegetation health, should talk with OS about embedding ML processes

By using machine learning, the capability now exists to monitor large areas of land, multiple times a year and at low cost. Any business focussing on land management, or important environmental areas such as land use, habitat loss, vegetation health, should talk with us to see whether ML processes could be integrated into your operations.

The above case studies focus on descriptive analytics which is using machine learning to gain a better picture of our past and current environment. Geospatial and AI (GeoAI) is evolving to include both predictive and prescriptive use cases with algorithms tasked with forecasting future changes and even suggesting solutions to specific business and policy challenges.   

 

indicative-maturity-scale

Ready to integrate ML / AI processes with your geospatial data? Contact us to learn more.

Newsroom team
By Newsroom team