We held the 54th and 55th Data Assimilation Seminars at RIKEN Center for Computational Science (R-CCS) on January 28.
The first talk was given by Dr. Alison Fowler from the University of Reading (UK), who spoke about observation error correlations and how we might apply data compression techniques that retain maximum information content of the observations, allowing us to use more observations in our data assimilation.
The second talk was given by Jemima Tabeart, also from the University of Reading. She spoke about available reconditioning methods that can be used to improve the condition number of correlated matrices. She found that she could improve the convergence by introducing correlated observation error covariance matrices with only small changes to retrieved variables.
We would like to thank Alison and Jemima for visiting us here at RIKEN R-CCS and giving two very interesting talks on observation error covariance matrix estimation – a very important topic in this era of Big Data!
For details, please see the following DA seminar page: