Adaptive Machine Learning: Pioneering Climate Modelling for a Sustainable Future
OPINION | By Dr Dhouha Kbaier, Senior Lecturer in Computing & Communications and Dr Ian Kenny, MSc Student and Post Doctoral Research Assistant , The Open University.
The Open University has a mission to generate impact-driven research to ensure a sustainable planet. Our contribution is through the intersection of two extraordinary fields: climate change and machine learning. We have created a model designed to adapt to the increasing rapidity of climate change.
Our starting point came from the realisation that accurate climate predictions are the backbone of informed decision-making, guiding everything from everyday weather forecasts to local responses to extreme weather events to long-term climate projections. However, traditional climate models face challenges in capturing recent events and minimising biases. In response, we have created an innovative approach that harnesses the power of adaptive machine learning, paving the way for real-time insights and dynamic decision-making in the face of our changing climate. The strength of machine learning is to identify patterns in data. Our aim is to produce an adaptive model able to respond to the changing climate more rapidly than traditional models.
Our mission is clear—to redefine climate modelling, empower decision-makers, and address urgent challenges such as food and water security and the consequences to human health posed by a changing climate. We want to design a model which is useful for local public service resource planning and for business continuity, enabling more resilience for society in the face of the effects of climate change.
By developing a model that adapts dynamically to smaller datasets, we aim not only to enhance climate predictions but also to inspire positive changes in policy, decision-making and behaviour for climate adaptation. The output of the model takes the form of probabilities as to an event occurring. However, the accuracy of these probabilities are a significant improvement over the current method of forecasting. This allows for better planning of future resource need in the short to medium-term: we are currently working on a five year adaptive cycle.
Accurate predictions are essential for preparing and mitigating the impact of extreme weather events, seasonal variations, and long-term climate shifts. We found that current traditional approaches often rely on very big datasets, which may not always be available for recent years or be sufficiently rich. Such data also faces the challenge of representing complex climate processes, which govern the increasingly complex interactions which take place in our atmosphere.
The model we developed is called RACC for Rapid Adaptive Climate Change. It takes a novel approach to tackle the challenge of generating accurate climate forecasts using smaller datasets to model and generate precise predictions.
The model integrates atmospheric and hydrospheric data so that the relationship between air temperature, humidity and barometric pressure can be measured, enabling a deeper understanding of the relationship between heat cycles . Through modelling this relationship it becomes possible to predict with a given likelihood what will happen next.
We expect this model to augment existing climate models rather than replace them, and we emphasise its potential to provide adaptive short-term forecasts for the evolving climate system.
In this era of rapid climate change, embracing innovative research enables us to refine our understanding and improve our ability to adapt to climate impacts. Our ongoing research represents a significant step forward in the realm of climate change modelling and forecasting and offers the potential to revolutionise such approaches.
In addition, the outputs from this research support climate adaptation planning at the local level, offering the opportunity for informed collective action to best prepare for climate impacts.
How can you get involved?
If you are a climate modeller or use climate models to inform decision-making and policy, or if you work in climate adaptation, you may wish to join our research group where you can:
- Help us build a multi-stakeholder advisory board;
- Join Workshops and Evidence Cafes to build your knowledge and networks;
- Inform the direction of the model development to meet your needs;
- Suggest research collaborations;
- Identify the research needs of different policy-makers and businesses on the frontline of climate adaptation;
- Engage in Citizen Science to enable us to gather local calibrated data to improve the model’s predictions;
- Provide your feedback on the practical usefulness of the potential outputs of the model.
At The Open University we are ‘open to people, places, methods and ideas’ and we are dedicated to ‘creating and sharing knowledge and learning to realise social and environmental justice.’
Image credit: Kyle Anthony Photography/Shutterstock.com
At The Open University we are ‘open to people, places, methods and ideas’ and we are dedicated to ‘creating and sharing knowledge and learning to realise social and environmental justice.’
To find out more about collaborating on our sustainability-related research, please visit Open Societal Challenges.
To find out more about how we can help you acquire the essential skills and confidence to respond to the nature and climate change crises, please visit:
Our free OpenLearn short courses ranging from 20 minutes to several hours;
Our apprenticeships
Our formal academic qualifications at undergraduate and postgraduate levels.
We have also specifically developed a microcredential – 100 hours of learning to help you take a big picture view and apply your learning to develop an action plan for sustainability in your organisation. Watch the course trailer here.
