Project Description:

To empower the Biodiversity Net Gain approach to development with precise satellite-driven products, it is vital to derive accurate and up-to-date habitat information. Our research focused on utilizing open multispectral and synthetic-aperture radar imagery to automate the detection of habitat and landscape changes, employing deep learning models to assign new semantic classes for altered areas. This research is essential for enhancing our comprehension of ecosystems to compute biodiversity metrics.  

The Intern will work as a part of the Earth Observation and Data Science Team in collaboration with colleagues, working on the product development to support the Biodiversity Net Gain. 

The specific part to which the Intern will contribute is the automatisation of the detection of habitat and landscape changes with the use of multispectral and synthetic-aperture radar imagery and state-of-the-art techniques for the imagery analysis with machine learning techniques. The Internship Host will provide the Intern with the necessary ancillary datasets, including the actual habitat layer, and access and training on using the cloud platform to process Earth Observation data. 

The research will focus on utilising open satellite data (especially Sentinel-2 and Sentinel-1), open geospatial technologies and programming for the imagery analysis. 

The general pipeline will be as follows: 

• Knowledge exchange on the product under development and peculiarities on the defining habitats and analysing their changes. Necessary capacity-building training and guidance. 

• Comprehensive literature review on comparative analysis of the different open habitat datasets, their creation and updating, and an analysis of the various change detection approaches. 

• Testing the approaches on the change detection with the satellite imagery and machine learning techniques. Implementing the selected approach on the national level dataset as a part of the Team. 

• Testing the open deep learning models on the land cover and habitat classification for the test samples within the changed areas. 

The exact workflow, scope, and outputs could be subject to change.  

• Find out more about SPIN and explore FAQs here.

Applicant Specification:

The successful Applicant should demonstrate a general understanding of the following fields with strong focus and interest in at least two of them: 

• Environmental Science. 

• Geospatial Science. 

• Earth Observation. 

• Data Engineering. 

• Programming. 

• Artificial Intelligence, Data Science and Machine Learning. 

• Software Development and Software Engineering. 

Minimum requirements is general understanding of the concepts of the following:

• Scientific and quantitative research basics. 

• Satellite Imagery analysis and processing.

• Programming and scripting.

• Machine Learning basics.

• Working with databases.

• Interest in learning Google Earth Engine.

  Process details:

• 8 weeks minimum fixed term contract to be agreed with successful candidate

• In-person Induction Event to be held on 6th May 2024, and an in-person/hybrid Showcase Event to be held at the W/C 24th June.

• 5 days holiday to be taken during the placement

This job description set outs the skills and experience we believe are needed to be able to do this job but, research also tells us women are much more likely than men to take this list of requirements as absolute and self-select out of the process. If you think you can deliver this role then we want to hear from you, regardless of the boxes you didn’t tick.