Satellites, statistics, and prediction: The science driving climate resilience
As the urgency to adapt to changing climate conditions grows, three research groups at KAUST are leading advances in science and technology that together offer a powerful framework for climate resilience. From space-based monitoring to statistical modelling of extreme events and cutting-edge hydrological forecasting, their work brings together critical disciplines needed to understand and respond to our evolving climate reality.
Observing changes from space
Matthew McCabe’s Hydrology, Agriculture and Land Observation (HALO) group is transforming how we monitor and respond to climate conditions using real-time space-based observation systems.
“Space-based Earth observation offers an incredible platform for capturing the state and condition of global processes,” says McCabe. “If we are to better understand and manage the Earth system, remote sensing will play a critical role in delivering insights and guidance.”
In 2023, the HALO team achieved a major milestone by launching its first satellite, a CubeSat designed for high-precision Earth observation in the arid Gulf region1. It is the first satellite to combine a GNSS Reflectometer, which tracks GPS signal reflections from Earth’s surface, and a hyperspectral sensor that maps land and sea to monitor vegetation, soil, agriculture, and ecosystem health.
By combining satellite data with modeling, the HALO team has created the most detailed estimates yet of irrigation water use and vegetation health, key information for tackling water scarcity. The team is also collaborating with the King Salman Royal Natural Reserve to support the management of rangeland ecosystems in northern Saudi Arabia.
“Turning Earth observation data into actionable intelligence requires considerable modelling and processing to deliver insights that operators and managers can actually use,” McCabe notes.
Much of the team’s work now focuses on developing a complete data processing and harmonization system. By applying big-data analytics and artificial intelligence (AI), they can generate actionable information and predictive models in near real time.
Chasing statistical extremes in a changing climate
While McCabe’s team observes from above, Raphael Huser’s Extreme Statistics (XSTAT) research group delves deep into the mathematical foundations of how we model and predict extreme weather and climate events. Their work addresses a fundamental challenge: predicting unprecedented events that could be more intense or cover a larger area than anything seen before.
“Risk assessment and planning in this context requires predicting extremes,” says Huser. “The statistical methods and models we develop must remain robust and reliable even when extrapolating beyond observed data. We create new models that have theoretical guarantees while maintaining flexibility in the ‘tail’ structure, which captures rare and extreme events,” he explains.
The team’s recent research has shed light on record-breaking extremes across the climate system2. Their analysis revealed alarming trends: between 2016 and 2024, record high temperatures on land occurred more than four times as often as would be expected without climate change. At the same time, record low temperatures happened half as often, while the frequency of daily maximum rainfall and monthly dryness records increased by 40% and 10%, respectively.
The XSTAT group is also advancing statistical methods to better understand extreme weather patterns across space and time. Their models show that in river basins like the Mississippi and Danube, extreme rainfall events are likely to become both more intense and more localized due to global warming3, which is a critical insight for future flood management. They have also developed hybrid statistical and deep-learning tools that forecast more frequent and severe wildfires in the Mediterranean region4.
To address the urgent need for sharper weather and climate predictions, KAUST researchers employ advanced methods: from real-time space-based observation systems to statistical models and comprehensive datasets.
Modeling compound climate extremes, events that occur together in space or time, is a major challenge, especially under shifting climate conditions. Traditional models struggle with non-stationarity and the complex interactions behind these extremes. Addressing this requires advanced modeling and statistical inference tools, high-performance computing, and closer collaboration between climate scientists, statisticians, and decision-makers.
“Combining state-of-the-art statistical, deep learning, and model-based tools, and the development of effective communication for weather and climate extremes and their impacts, along with appropriate measures of uncertainty, are key points to improve resilience and better prepare for future extreme events,” Huser says.
Predicting hydrological extremes
Completing this interdisciplinary triad, Hylke Beck’s Hydro-climatic Extremes (HYEX) Group focuses on translating observations and statistics into predictive capabilities. In 2023, they released a comprehensive high-resolution global map of climate zone changes from 1900 to 2100 across seven socioeconomic scenarios, revealing dramatic historical and predicted future transformations in vegetation and ecosystems5.
“Our modelling showed that under even a moderate greenhouse gas emissions scenario, an area of approximately 2.6 million square kilometers is expected to transition from a frozen polar to seasonal cold climate,” says Beck.
Beck’s team has developed one of the most advanced global precipitation datasets to date: the Multi-Source Weighted-Ensemble Precipitation (MSWEP) dataset. It combines satellite data, climate models, and ground-based observations to deliver highly accurate rainfall estimates across regions worldwide.
MSWEP supports a wide range of users: from water managers modeling streamflow, to meteorologists testing weather models, to reinsurance companies refining risk assessments. It’s also been used to analyze extreme events, such as the UAE floods in late 2024, to explore possible links to climate change.
“The findings of that study showed climate change increased the intensity of the floods by 10–40%,” says Beck.
The latest version of MSWEP, two years in the making, uses advanced machine learning models to better combine satellite, model, and ground data, further improving rainfall accuracy. “This sophisticated blending is already yielding key insights,” says Beck. For example, recent research6 from an international consortium led by Beck shows that global drought severity has risen by about 40% over the past 120 years, driven largely by heat-induced increases in atmospheric evaporative demand.
Still, Beck stresses that technical progress isn’t enough; action and communication are just as critical. He points to tragic cases where data existed but wasn’t used, warning that “authorities must resist pressure to approve construction in vulnerable areas, and that warnings are useless if they don’t reach people or trigger action.”
The convergence of these three research programs at KAUST – from observation to analysis to prediction – creates a powerful platform to support climate adaptation and resilience, especially in arid regions like Saudi Arabia. As climate extremes grow more severe, this kind of interdisciplinary collaboration is not just valuable but essential for adapting to a changing world.
References
1. McCabe, M. F., Aragon, B., Houborg, R., & Mascaro, J. CubeSats in hydrology: Ultrahigh-resolution insights into vegetation dynamics and terrestrial evaporation. Water Resources Research, 53, 10017–10024 (2017). https://doi.org/10.1002/2017WR022240
2. Fischer, E.M., Bador, M., Huser, R. et al. Record-breaking extremes in a warming climate. Nature Reviews Earth and Environment 6, 456–470 (2025). https://doi.org/10.1038/s43017-025-00681-y
3. Zhong, P., Brunner, M., Opitz, T., & Huser, R. Spatial modeling and future projection of extreme precipitation extents. Journal of the American Statistical Association 120(549), 80–95 (2024). https://doi.org/10.1080/01621459.2024.2408045
4. Richards, J., Huser, R., Bevacqua, E., & Zscheischler, J. (2023), Insights into the drivers and spatio-temporal trends of extreme Mediterranean wildfires with statistical deep-learning, Artificial Intelligence for the Earth Systems 2, e220095
5. Beck, H.E., McVicar, T.R., Vergopolan, N. et al. High-resolution (1 km) Köppen-Geiger maps for 1901–2099 based on constrained CMIP6 projections. Scientific Data 10, 724 (2023). https://doi.org/10.1080/01621459.2024.2408045
6. Gebrechorkos, S.H., Sheffield, J., Vicente-Serrano, S.M. et al. Warming accelerates global drought severity. Nature 642, 628–635 (2025). https://doi.org/10.1038/s41586-025-09047-2