[17] Observational issues 2

[17-3] March 2, 11:40-12:00

On the interaction of observation and a-priori error correlations in data assimilation

Alison Fowler (University of Reading), Sarah Dance (University of Reading) and Joanne Waller (University of Reading)


The inclusion of observation error correlations (OECs) in data assimilation (DA) is currently receiving much attention. It is expected that by allowing for a more accurate description of the observation errors, OECs will result in a more optimal DA system and hence reduce the analysis errors. Accurately accounting for OECs will also allow for more dense observations to be assimilated which could be beneficial for high resolution and high impact forecasting. However, we will show that the impact of observations with correlated errors depends not only on the characteristics of the OECs but also on those of the a-priori error correlations (BECs) and the observation operator (H). Exactly how the OECs interact with the BECs and H will be illustrated using a series of simple 2 variable experiments and a variety of metrics, including the analysis error correlations, the sensitivity of the analysis to the observations and information content. The results from these experiments will then be used to inform the optimal observation density for a variety of observation types.

  Presentation file: 17_3_A.Fowler.pdf