nmarzz/MCRN-DA

This project involves developing reduced physical and data models and algorithms for data assimilation. The goal of the project is to address computational challenges in data assimilation of dimensionality and non-Gaussian behavior for model problems and relevant small to medium scale problems. This includes a models from many different possible areas: atmosphere, ocean, combined atmosphere/ocean, ice sheet, glacier, hurricane, ENSO, polar vortex, ecological models, etc.. The basic idea is to create computational conceptual models, both physical and data models, and combine these with standard data assimilation techniques and new techniques developed to take advantage of the structure of these conceptual models. Among the data assimilation techniques to be considered are projected particle filters. Focus is on application to problems with bimodal or multimodal behavior which ties in with work on tipping points. Another emphasis is on applying these techniques to higher dimensional problems to create lower dimensional computational conceptual models. The project involves employing, developing, and applying projected data assimilation techniques, in particular projected particle filters, using the framework developed in (Maclean, VV 2019), while focusing on the use of different state space and observation space projections for increasingly high dimensional models.

MATLAB