FEIT Research Project Database

Data assimilation

Project Leader: Jonathan Manton
Collaborators: Professor Craig Bishop (Earth Sciences)
Primary Contact: Jonathan Manton (jmanton@unimelb.edu.au)
Keywords: signal processing
Disciplines: Electrical & Electronic Engineering
Research Centre: Nonlinear Signal Processing Lab

Currently, one of the main impediments to enhanced weather forecasts is data assimilation, the challenge of determining the correct initial conditions for the numerical weather model. From a signal processing perspective, data assimilation is a statistical filtering problem. The key challenge though is the sheer size of the data; it is computationally intractable to implement all but the most basic of existing filters given that computing tomorrow's forecast must take significantly less than a day to be useful. By starting with simpler but representative models (such as the shallow-water equations), this project will develop novel filtering techniques for data assimilation that improve upon linear filters yet remain computationally tractable.