Our research focuses on developing a seamless framework for weather and climate prediction, bridging short-term forecasts with long-term climate projections. We investigate the dynamical processes governing atmospheric and oceanic variability, addressing both initial value problems for short-term predictions and forced boundary condition problems for long-term climate projections.
We study the predictability of weather and climate on timescales from days to seasons, improving early warning systems and seasonal outlooks.
Our work explores the mechanisms driving interannual to decadal changes, refining climate models to capture regional variability.
We analyze how external forcings shape future climate, producing high-resolution projections to assess regional climate impacts.
By integrating numerical modeling, observational data, and machine learning techniques, our research enhances the accuracy of climate predictions and supports informed decision-making in a changing environment.
Together, these six pillars form the backbone of our research. By combining remote sensing, physics, analytics, AI, HPC, and autonomous systems, we enhance the accuracy of climate predictions, improve early warning capabilities, and provide robust knowledge to inform sustainable decision-making in the face of climate variability and change.
We develop and deploy robotic platforms, such as autonomous underwater vehicles and aerial drones, equipped with state-of-the-art sensor technologies. These systems extend our ability to monitor and explore environments that are difficult to access, while providing adaptive, automated data collection. Their integration into the digital twin strengthens its real-time accuracy and enriches its observational depth.
Meeting the scale and complexity of the digital twin requires the computational power of KAUST’s advanced HPC infrastructure. This resource enables us to run highly detailed simulations and couple multiple processes—physics, data streams, and AI models—seamlessly and efficiently. The result is the capacity to investigate unprecedented scales of environmental variability and project their impacts with precision.
Machine learning and AI algorithms empower us to handle the immense volume and diversity of environmental data. By detecting hidden patterns, generating predictive insights, and continuously refining models, AI enhances the responsiveness and adaptability of the digital twin. This ensures that the system not only reflects current conditions but also anticipates emerging trends with greater accuracy.
Our work is underpinned by state-of-the-art models grounded in physical laws to simulate key environmental processes. In situations where observational data are scarce, physics-based modeling plays a crucial role in filling critical gaps, ensuring continuity and scientific rigor. These models enable us to explore interactions among the atmosphere, oceans, and land, bringing the digital twin to life with scenarios that are not only realistic but also validated against established scientific principles.
We draw on historical archives and real-time observations from a wide range of sources—satellites, ground-based instruments, and oceanic sensors. These data streams continuously feed into the digital twin, ensuring it is informed by the latest field observations and offering a comprehensive view of environmental conditions across scales. This integration provides the foundation for reliable monitoring and situational awareness.
Our work is supported by three key collaborators: the Climate Change Center, the national hub for climate science and operational forecasting; Aramco, leveraging advanced climate and marine intelligence for safe and resilient operations; and NEOM, pioneering high-resolution climate tools and datasets for sustainable urban and coastal development.
See how each collaboration drives impact:
