Theory and Practice of Data Assimilation in Ocean Modeling

摘要

The long-range goal of this project is to form the best picture of the ocean as an evolving system based on data assimilation, i.e., the construction of a composite estimate of the state of the ocean based on a combination of observed data with computational model output, and to 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 taken together, one can 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 the technical challenges of 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.