Climate change is currently the most significant threat to our way of life on earth. No one is immune to its effects. A study based at our Singapore campus is exploring how AI can help to predict future weather patterns.
Unprecedented flooding, heatwaves, droughts and wildfires are occurring all over our planet, and it’s critical that key policy-makers and stakeholders always have the most accurate information available. This enables them to make better-informed decisions on tackling climate change.
However, acquiring this information and determining what our future climate will look like is an enormously complex computational task, requiring the building and running of Earth System Models (ESMs) that encompass the various components and processes that occur on our incredible planet. We also need to be able to run many different scenarios to understand how our future actions could affect the trajectory of climate change.
Driving change from Singapore
Though no country on Earth will escape its effects, Singapore will be particularly impacted by future climate conditions.
As a low-lying island, 30% of Singapore is less than 5m above mean sea level. Damage caused by a rising sea level, increased rainfall, and higher temperatures would be catastrophic, and present significant challenges to the management of the country’s water resources. Periods of drought can affect the reliability of Singapore’s water supply, while sudden episodes of intense rainfall could overwhelm its drainage system and lead to flash floods.
Singapore is also particularly vulnerable to the effects of climate change in other countries. As Singapore imports more than 90% of its food, intense storms, flooding, and prolonged droughts are a constant threat to its access to consumables, and even wider global food production.
Our partnership with NVIDIA
A new PhD project based at Newcastle’s Singapore campus is working with NVIDIA – one of the largest global technology companies in the world – to improve the accuracy and performance of climate models. The project is funded by the Singapore Economic Development Board and NVIDIA.
Founded in 1993, NVIDIA pioneers accelerated supercomputing and artificial intelligence to tackle challenges that otherwise can’t be solved. NVIDIA’s invention of the GPU in 1999 sparked the growth of the PC gaming market, and has redefined modern computer graphics, high performance computing, and ignited the era of modern AI. Its work in AI and computer graphics is transforming global industries valued at more than $100 trillion, from gaming to healthcare to transportation, and is profoundly impacting society.
The quest for answers
The sheer scale of our planet and the numerous processes that occur make it computationally impracticable to simulate everything numerically. Current models are also limited in resolution, and can only directly resolve large-scale features at around 100km in size.
Smaller features are approximated through sub-grid scale (SGS) parameterisations of the physical processes. These estimates lead to uncertainties in the model, and limit our confidence in our predictions for climate change. It is thought that we won’t have sufficient computational capability for directly resolving these important features until the year 2060 – and that’s too far away for the answers we need today to impact important policy changes.
Artificial Intelligence techniques provide us with an alternative solution.
We can use AI to help our climate models with techniques such as AI augmentation, data-driven prediction, and emulation. Our PhD project focuses on applying AI emulation to the complex task of modelling atmospheric chemistry – a fairly recent addition to the world of climate models.
AI and atmospheric chemistry in climate models
Atmospheric chemistry plays an important role in determining the impact of climate change.
Atmospheric chemistry is a branch of atmospheric science focussing on the chemical processes within Earth's atmosphere. It draws on environmental chemistry, physics, meteorology, computer modelling, oceanography, geology and volcanology and other disciplines.
Our climate is changing because earth’s energy budget is unbalanced, with the planet receiving more energy than it emits. This is mostly caused by pollutant gases (greenhouse gases) and clouds. Clouds provide the greatest amount of uncertainty in future climate predictions, and the interaction between clouds and pollutants is complex and computationally challenging. This is why understanding atmospheric chemistry is vital to making climate change predictions.
A future of accurate climate predictions
By applying surrogate AI models to these complex chemical processes and the interactions between aerosols and clouds, the study hopes to make it possible to run these models much faster than is possible today with numerical simulation alone. This will then pave the way for faster and more accurate predictions of future climate conditions and enable more timely responses to future ‘what if’ scenarios.
The study is comprised of a team of cross-disciplinary specialists, including;
- Dr Cheng Chin – Reader in Intelligent Systems Modelling and Design, brings expertise in electronics and artificial intelligence
- Dr Simon See – FIET, SMIEE, is the industry supervisor for the project
- Mr Jeff Adie – Principal Scientist and PhD candidate for the project.
The project continues to progress successfully, and the team are continually exploring further opportunities for collaborative research projects between Newcastle University in Singapore and NVIDIA.
Keen to find out more about how our work is helping to make the world a better place to live in?