CIRA » Home » Research » Dimension Reduction in Dynamical Systems
Ensemble data assimilation and prediction, with high-dimensional weather and climate systems, potentially requires large numbers of ensembles, eventually creating a limit for feasible applications. In principle, there are two approaches: development of reduced-order models, or the covariance localization. The former approach has not been used in ensemble data assimilation, being mostly connected with modeling development. In ensemble data assimilation the latter approach is generally used.
We are interested in both approaches and their further development. The dimension reduction is closely related to the degrees of freedom of the system, information measures, and to the number of ensembles. Its understanding will shed light on the most important adverse characteristic of ensemble data assimilation - the insufficient number of degrees of freedom.
The dimensionality of nonlinear dynamical system is closely related to the chaotic nature of weather and climate systems, and ensemble data assimilation and prediction should also be interpreted in the context of chaotic dynamics.
This research topic is a component of other research topics
- Model Error and Bias
- Predictability
- Applications with Complex Models
- Applications with Simple Modles
- Basic Development of Ensemble Data Assimilation
and research projects
- Weak constraint approach to ensemble data assimilation: Application to microwave precipitation observations
- Research and development for GOES-R risk reduction
- Ensemble Kalman filtering for Army-scale meteorology