Theory and Practice of Data Assimilation in Ocean Modeling
Abstract
The long-range goal of this project is to combine computational models with observational data to form the best picture of the ocean as an evolving system, and use that picture to understand the physical processes that govern the ocean's behavior. Oceanic observations are sparse and models are limited in accuracy, but judiciously constructed combinations of data and model output have the potential to form a quantitative description of the state of the ocean that is superior to any based on either models or data alone. Along with the goals of analysis and prediction, we seek reliable estimates of the errors in our results. We expect our results to have implications beyond data assimilation. In particular, we believe this research will lead to enhanced understanding of the implications of nonlinearity and randomness for predictability of the ocean and atmosphere. In keeping with our goal of providing reliable error estimates for our data assimilation products, we seek to develop efficient methods for estimating useful statistical measures of errors in stochastic forecast models. Since the probability density functions (PDFs) of nonlinear stochastic models are not, in general, Gaussian, we must find methods for forecast evaluation based on information about the particular PDF generated by the model.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 30, 2007
- Accession Number
- ADA573204
Entities
People
- Robert N. Miller
Organizations
- Oregon State University