POST (Parliamentary Office of Science and Technology)
![]() |
Remote Sensing and Machine Learning
There is increasing interest in using machine learning to automatically analyse remote sensing data and increase our understanding of complex environmental systems. While there are benefits from this approach, there are also some barriers to its use. This POSTnote examines the value of these approaches, and the technical and ethical challenges for wider implementation.
Download the full report
-
Remote sensing and machine learning (PDF, 376 KB)
Environmental remote sensing involves the use of satellites and other air-borne instruments to collect data about the environment. Substantial quantities of data are produced in this way and environmental remote sensing is now considered an area of Big Data. Experts are using artificial intelligence (AI) tools such as machine learning for more efficient data analysis of such data.
Machine learning algorithms allow a system to learn and improve from data and experience without being specifically programmed, reducing the level of human intervention. This data-driven approach means valuable information about a natural phenomenon can be extracted from the data alone. This has benefits such as being able to manage more complex environmental data but has challenges such as data accessibility.
Key Points:
- Machine learning with remote sensing can help to improve predictions about the behaviour of environmental systems, improve the automation of data analysis, lead to a better management of resources and the discovery of new insights from complex data sets.
- Applications at this interface include improved weather forecasting, flood and drought prediction, precision agriculture, managing forests and in marine conservation and coastal clean-up projects.
- Wider implementation for remote sensing is limited by the availability of accessible, and representative datasets for training the machine learning algorithm. Specific challenges include: the availability of Analysis Ready Data (ARD) that requires expertise, time and computational power to prepare; the demands on storage, transfer and processing of large data sets; and, the demands on having an accurate and well-developed training data set.
- Inherent to all remotely sensed data used to train and test the models is a level of uncertainty and error including error introduced by a human analyst. Best practice guidelines vary and there is an overall lack of regulation by Government or relevant public bodies. Uncertainties may also be poorly understood by stakeholders.
- Machine learning models are sometimes referred to as ‘black box’, implying that it is not easy to determine how an algorithm made a decision. For a public sector organisation, where any aspect of a decision is opaque, that raises issues of accountability and having the authority to make a decision.
- As machine learning becomes of greater value to the UK economy, challenges could be addressed by stakeholders such as NGO’s, academics and commercial EO companies working with Government to establish good practice codes, standards and regulatory frameworks.
Acknowledgements
POSTnotes are based on literature reviews and interviews with a range of stakeholders and are externally peer reviewed. POST would like to thank interviewees and peer reviewers for kindly giving up their time during the preparation of this briefing, including:
Alberto Arribas, Met Office
Heiko Baltzer, University of Leicester
John Bloomfield, British Geological Survey
Julie Bremner, University of East Anglia & CEFAS*
Rene Breton, University of Manchester
Sue Chadwick, Pinsent Masons LLP*
Michael Cross, Rezatec*
Timothy Darlington, Met Office
Tony Dolphin, CEFAS*
Joseph Fennell, University of Manchester
Karen Frake, Natural Scotland*
Tanvir Islam, NASA Jet Propulsion Laboratory
Tom Jones, Satellite Applications Catapult
Gwawr Jones, JNCC*
Alexandra Kilcoyne, Natural England
Peter Kohler, CEFAS
Stefan Leutenegger, Imperial College London
Paula Lightfoot, JNCC*
Encarni Medina-Lopez, University of Edinburgh
Paul Monks, University of Leicester
Boguslaw Obara, Durham University*
Rami Qahwaji, University of Bradford
John Remedios, NCEO and University of Leicester
Cristian Rossi, Satellite Applications Catapult
Edward Salakpi, University of Sussex
Anna Scaife, University of Manchester
Ivan Tyukin, University of Leicester
Hong Wei, University of Reading
Daniel Wicks, Satellite Applications Catapult
*denotes people and organisations who acted as external reviewers of the briefing.
- POSTnote POST-PN-0628
- Authors: Berivan Esen, Jonathan Wentworth
- Topics: Digital tech, Environment, Food security
Download the full report
- Remote sensing and machine learning (PDF, 376 KB)
Original article link: https://post.parliament.uk/research-briefings/post-pn-0628/
![]() |
Academic Fellowships | Upcoming work | POST Publications | ||
|