When it comes to tackling the global climate crisis, data can be one of the most powerful tools we have. Yet, the sheer volume of data needed to accurately predict our changing climate is immense, requiring supercomputers that can process quintillions of calculations per second. Enter Dr Zubair Khalid, an associate professor at the Syed Babar Ali School of Science and Engineering (SBASSE) at LUMS, who recently became the first Pakistani to win the prestigious ACM Gordon Bell Prize for Climate Modelling. Along with an international team of researchers, Dr Khalid’s success points to the exciting ways in which advanced computing and AI can help not only the world at large but also the communities most vulnerable to climate change’s devastating consequences at home in Pakistan.

Awarded by the Association for Computing Machinery (ACM), the Gordon Bell Prize is sometimes called the “Nobel Prize of Supercomputing.” It recognises landmark achievements in high-performance computing (HPC) that push the boundaries of what modern computers can achieve. In November 2024, Dr Khalid and 11 other collaborators received the prize for their work titled ’Boosting Earth System Model Outputs and Saving PetaBytes in Their Storage Using Exascale Climate Emulators’.

At a time when rising temperatures, catastrophic floods and intensifying smog have become the norm, robust climate models are essential for predicting and mitigating local and global climate impacts. “I have been working on spatial data analysis for about 10 to 12 years. The project began in 2022 when I visited KAUST (King Abdullah University of Science and Technology in Thuwal, Saudi Arabia) during the summer. We discussed the increasing demand for climate data, which typically comes from running physics-based (partial differential) equations on supercomputers to achieve high-resolution outputs. If we need only low-resolution data, it is possible to use desktop computers, but for higher resolution – in space and time – we need large-scale computing resources.

Traditional models require tremendous computational resources and produce massive datasets. By contrast, the team’s ground-breaking exascale climate emulator dramatically reduces storage needs while enhancing the level of detail and accuracy in climate simulations. “Rather than taking an existing algorithm and simply running it on a supercomputer, we developed the underlying science – how to generate and store high-resolution climate data efficiently – and then scaled those approaches.” They asked, “How can we address the immense compute power and storage requirements for climate simulations?” That resulted in a strategy to save petabytes of storage while still generating accurate, high-resolution climate data.

High-Resolution Climate Data – On Demand

Dr Khalid explains that their emulator relies on AI and HPC to do something both innovative and essential: produce high-resolution climate data on demand without having to store every single data point. Using sophisticated algorithms (including the spherical harmonic transform and Cholesky factorisation), the group can emulate the massive amounts of data produced by ultra-high-resolution climate models. The result is a system that can capture small-scale phenomena – like the variation of smog in a single district or an intense heatwave’s precise movement across a region – while saving petabytes of storage.

“Many extreme events, like multi-week smog episodes or major heat waves, are influenced by global phenomena such as El Niño. The point is that everything is interconnected – what happens in one region will eventually affect another.”

For a country like Pakistan, where extreme climate events – floods, droughts and heatwaves – are alarmingly common, the ability to quickly and accurately model local impacts is a game-changer. The model can be localised for Pakistan, potentially offering hourly forecasts at resolutions as fine as a few kilometres. Such precision is crucial for policymakers, local governments and NGOs seeking to protect vulnerable communities.

Bridging Gaps in Climate Research and AI

“Pakistan is among the most vulnerable countries affected by climate change,” says Dr Khalid. Yet, despite the urgency, HPC resources remain heavily concentrated in wealthier nations. He acknowledges this obstacle but believes that investing in fundamental science education and local HPC capacity can help bridge that gap. “I would again go back to the fundamentals. At universities, in academia – even high school – there is a need to realise that our kids and our students go after fundamental knowledge, building foundations.”

“If there is a gap in the fundamentals, and if they don’t bridge that gap while they are at high school or university, it will remain forever.”

He points to collaborative efforts with the Punjab government, where high-resolution climate data is being used to analyse pollution sources and strategise mitigation. The success of this project, he emphasises, shows what is possible when multidisciplinary teams – combining climate science, statistics, AI and HPC – work together. In his view, strong fundamentals make these collaborations more effective.

A Historic Win for Pakistani Academia

Dr Khalid’s Gordon Bell Prize is a milestone for Pakistani academia. It demonstrates that fundamental, home-grown research can have a tangible impact on complex and global problems. He aims to foster more talent in Pakistan by encouraging students and researchers to build strong math and computational foundations early on. “I have supervised about 10 PhD students and see them as my biggest asset. Watching them grow is immensely rewarding. If you have solid fundamentals,” he explains, “you can apply them to any domain – whether it’s climate change, finance or human rights.”

The distinction between learning skills and learning foundational knowledge is critical. “It’s a trade-off between knowledge and skill. For sure, we need to develop skills. But universities should provide a platform where we don’t only teach skills, but we transfer the foundation of fundamental knowledge.” Such foundational research, he points out, is often undervalued in favour of direct applications, but it must be done. “At times, we hear that you don’t need to do fundamental research – just do the applied research so that people can get great benefit. But this foundational research has to be done by someone, right?”

According to Dr Khalid, bridging technology, data and on-the-ground reality is key to addressing challenges such as heat stress, water scarcity and pollution in regions across the country.

This fundamental approach is exactly what Dr Khalid and his colleagues teach at LUMS, where they have set up dedicated research centres like the Center for Urban Informatics, Technology and Policy (CiTy). “A few years ago, my colleagues and I started CiTy at LUMS, aimed at using data and AI to address urban challenges – air pollution, water distribution, population shifts and more. It’s a blend of fundamental and applied research that showcases how advanced computing can help solve pressing societal problems.”

Looking Ahead

For those of us worried about how climate change affects our communities – or curious about the intersection of AI, HPC and scientific discovery – Dr Khalid’s work shows a path forward. By coupling advanced computing with locally sourced data, these new models can help everyone from farmers in rural Pakistan to urban planners in Karachi better prepare for and adapt to the evolving climate.

Dr Khalid’s historic recognition strongly encourages investing in big ideas and big computing to address global challenges. In a country often short on research funds and resources, it’s an inspiring testament to what’s possible with passion, collaboration and unwavering dedication to scientific progress.

As Pakistan grapples with environmental crises – smog, deadly floods and chronic water scarcity – the real-world impact of Dr Khalid’s exascale emulator could be profound. In the near future, city-wide pollution control policies or timely flood warnings may well hinge on the insights gleaned from high-resolution climate simulations, helping protect lives and livelihoods across the nation.