Large-scale modelling and data assimilation

Modelling physical processes in the hydrosphere is one of the core research areas of the Water and Environment group in IBM Research – Ireland. The group's interests span from theoretical aspects of large-scale modelling to their applications in hydrodynamics and water quality, flood monitoring and forecasting, waves and oil and gas leak/spills prediction, among others.

Approach

The group's approach is holistic and systemic in nature, solving research problems by improving and creating techniques that integrate data, models, metamodels and their computational efficiency and scalability on high-performance platforms.

  1. Develop robust numerical methods and mathematical tools to understand physical phenomena.
  2. Develop an improved understanding of physical processes by combining physics and data analysis through the results of the numerical simulations.
  3. Integrate/Improve these methods to realistic settings by including aspects such as data assimilation, inverse parameter estimation and data analysis.
Sean McKenna Emanuele Ragnoli Fearghal O Donncha Noreen O Brien Seshu Tyrupathi Sergiy Zhuk
Beat Busser Albert Akhriev

Numerical methods

This group concentrates on developing novel numerical methods to solve the mathematical equations that define the physical process. A wide range of methodologies from finite differences, finite elements, finite volumes, spectral methods and discontinuous Galerkin methods are investigated, improved and applied. Novel automatic adaptive meshing and subgrid approaches are investigated and developed.

High-performance computing

This group focuses on developing and investigating high-performance computing techniques that range from MPI to Open MP techniques, combining them in hybrid approaches to better scale and exploit traditional and new high-performance computing platforms.

Data assimilation

This group focuses on improving current linear and nonlinear filtering techniques with heterogeneous data sources. These are integrated in high-performance computing platforms for real-time integration of models and sensors data within operational settings.  Novel research is being conducted to investigate and develop new approaches to nonlinear filtering in realistic settings.