The 2020 Geography in Government Awards opened for nominations at the beginning of the year, seeking examples of excellence in geography across the range of disciplines and organisations. This is the second year we have held the awards and all the judges were pleased by the outstanding quality and range of examples the profession can offer. Reading the nominations it was very clear that geography in government is a broad profession with excellence across the public sector and a profession which is full of dedicated and talented individuals.
Although we’re unable to host the award ceremony in person this year, I would like to take this opportunity to share the short listed nominations for each category. We are planning to run a virtual ceremony in the near future where the category winners will be announced, along with the overall winner. This overall winner is invited to the hugely prestigious Royal Geographical Society Awards, recognising the impact and importance of geography across the public sector.
So, here are the shortlisted nominations for the category of "Advancing geospatial data science", along with a short description of the work in alphabetical order.
In the Thames region, more Natural Flood Management (NFM) is needed to help tackle the climate emergency. By emulating natural processes such as slowing surface runoff and increasing storage, NFM can increase the landscape’s capacity to mitigate flooding while benefiting important habitats and wildlife. Spatial analysis techniques in Geographic Information Systems (GIS) were used to automate the correlation of 14 data sets representing landscape attributes. At the catchment scale, this automated approach provides a more objective result than manual interpretation.
ONS Geography Team
This nomination covers work undertaken by the Office for National Statistics Geography Team in 2019 in contributing to the development of a Discrete Global Grid System. This is a standard created by the Open Geospatial Consortium to spatially reference geospatial information globally. Leading on research for a Discrete Global Grid System allowed the British Isles to be mapped onto grids using various cell shapes and sizes covering the entire world to solve some of the problems faced in presenting geospatial data in a world context.
Data Science, Engineering and Remote Sensing Teams, UK Hydrographic Office
The team designed a deep learning model for image segmentation using a convolutional neural network, trained to detect mangrove. The Data Science and Remote Sensing teams worked collaboratively to create a labelled dataset of 40 Sentinel-2 images used to train the model. Overall, the trained model achieved a balanced accuracy of 82% with well controlled regional differences in performance, demonstrating that the model is geo-generalised. The Data Engineering team then developed a serverless pipeline to deploy the model and make global predictions of mangrove extent.