Data Assimilation Meetings at Reading
Date | Meeting type | Speakers |
31 October 2012 | DARC seminar |
Phil Browne (UoR) Generating 3D adapted grids for data assimilation Following the successful implementation of mesh adaptation within vertical levels of the UK4 model for data assimilation, the University of Bath and the Met Office undertook a Knowledge Transfer project to investigate the feasibility of fully 3D adaptive grid generation for the same purpose. Based on a given monitor function, an r-adaptive method will seek to move the grid points so that they are dense in regions where the monitor function is high, and sparse where the monitor function is low. To date, very little work on 3D r-adaptive methods exists, and what does exist is computationally expensive. The University of Bath has done a great deal of work on 2D r-adaptive methods by solving the Parabolic Monge-Ampere (PMA) equation which has proved robust and computationally cheap. In this talk we will show how the 2D PMA equation extends into 3D with some interesting model examples. Using real forecast data, we will show how we have defined a robust monitor function which behaves well in different scenarios. Finally we will show the resulting grids generated by the 3D PMA method and look its operational feasibility which shows great potential for improving the quality of the data assimilation. |