Difference between revisions of "Data Assimilation For Fluid Dynamic Models"
m (1 revision) |
|||
| (One intermediate revision by the same user not shown) | |||
| Line 1: | Line 1: | ||
| − | + | '''Mentor:''' [http://www.marquette.edu/mscs/facstaff-spiller.shtml Elaine Spiller] | |
| − | + | '''Approach:''' Implement and evaluate numerical methods to solve stochastic differential equations modeling | |
| − | + | ||
| − | Approach: Implement and evaluate numerical methods to solve stochastic differential equations modeling | + | |
atmospheric and oceanographic fluid flows. | atmospheric and oceanographic fluid flows. | ||
| − | Summary: Data assimilation is a broad term for techniques that combine noisy observations with dynamic | + | '''Summary:''' Data assimilation is a broad term for techniques that combine noisy observations with dynamic |
model based predictions. Applications vary widely; some interesting examples are weather prediction/storm | model based predictions. Applications vary widely; some interesting examples are weather prediction/storm | ||
tracking, ecological population studies, and real-time traffic flow. Many data assimilation schemes are based | tracking, ecological population studies, and real-time traffic flow. Many data assimilation schemes are based | ||
Latest revision as of 02:18, 20 January 2017
Mentor: Elaine Spiller
Approach: Implement and evaluate numerical methods to solve stochastic differential equations modeling atmospheric and oceanographic fluid flows.
Summary: Data assimilation is a broad term for techniques that combine noisy observations with dynamic model based predictions. Applications vary widely; some interesting examples are weather prediction/storm tracking, ecological population studies, and real-time traffic flow. Many data assimilation schemes are based on assumptions that models are approximately linear and that the uncertainty of a system follows a Gaussian distribution, both of which are often invalid assumptions. This project is concerned with applications where data is (spatially) sparse and obtaining data on a grid is challenging – as is typical of atmospheric and oceanographic problems. This is an exciting area of study since the applications are diverse and important to help us understand the current and future state of our planet. Scientifically, it is important to develop methods that can deal with the available data (often from weather balloons, ocean drifters or ocean gliders in cases of environmental science). Since data in such problems comes to us in the Lagrangian frame, or semi-Lagrangian frame in the case of gliders, it is useful to establish data assimilation methods that work in different frames of reference. The ultimate scientific goal is to establish methods in the Lagrangian frame that can also naturally deal with nonlinearity. The project will combine techniques from dynamical systems theory with traditional, statistically based data assimilation methods to accomplish this task. Students involved in this project will learn about data assimilation and fluid dynamic models, implement numerical methods to solve stochastic differential equations, learn about and implement filtering methods (i.e., updating models with data), and learn about flow behavior from a dynamical systems perspective.
