Predicting the Earth's future climate
The Earth's climate dynamics involve complex multi-physics multiscale processes that can only be predicted through a combination of data and large-scale physics-based models.Learn more ↓
Oden Institute Computational Research in Ice and Oceans Group
Reconstructing the past and predicting the Earth's future climate
Global climate models represent various components of the coupled Earth system, including the ocean, atmosphere, sea ice and ice sheets. These models encapsulate the fundamental laws of physics to simulate the dynamics of geophysical flows on the rotating planet, and the physical interactions and feedbacks between its components. For example, these physics-based models embody the equations of motion that govern the ocean circulation — conservation of mass, momentum, energy, and salt – and its interaction with overlaying atmosphere and sea ice.
In-situ ocean sensors and satellite remote sensing missions provide valuable data to inform construction, calibration, and initialization of these models. However, the data are sparse, inhomogeneously distributed in space and time, and heterogeneous in terms of state variables being measured, providing only limited peeks into time-evolving, full-depth ocean state. They are also very expensive to acquire. High-consequence decisions about where and how to deploy new sensors must be informed by numerical simulations that use all available physical knowledge to fill data gaps, provide a coherent, dynamically and kinematically consistent picture of the oceanic state, and account for the different sources of uncertainties. The multi-physics multiscale dynamics have complex interactions, and the available data cannot by themselves reveal key climate indices needed to understand the processes at play, let alone issue skillful predictions. Computational science provides powerful theoretical frameworks and numerical algorithms to this end, that have yet to be fully brought to bear in climate modeling.