Premium
On the sensitivity of a 4D‐Var analysis system to satellite observations located at different times within the assimilation window
Author(s) -
McNally Anthony P.
Publication year - 2019
Publication title -
quarterly journal of the royal meteorological society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.744
H-Index - 143
eISSN - 1477-870X
pISSN - 0035-9009
DOI - 10.1002/qj.3596
Subject(s) - data assimilation , meteorology , computer science , environmental science , sensitivity (control systems) , satellite , advection , window (computing) , sliding window protocol , assimilation (phonology) , geography , aerospace engineering , linguistics , philosophy , physics , electronic engineering , engineering , thermodynamics , operating system
This study quantifies the extent to which the ECMWF 4D‐Var displays differential (heightened) sensitivity to observations located near the end of the 12‐hr assimilation time window compared to observations located near the start of the window. Using dedicated satellite data denial experiments, it is shown that the lattermost 3 hr of observations are significantly more influential on the quality of the assimilation and forecasting system than the first 3 hr of data. Furthermore, it is found that the last 3 hr of data even outperforms the 6 hr of data (i.e. twice the number of observations) located in the first half of the window. The heightened importance of late window data is discussed in terms of these measurements being our most up‐to‐date information on the atmosphere, but also their ability to provide additional dynamical information to the assimilation system via feature advection wind tracing. The implications of this sensitivity are discussed. Firstly, it leads to the existence of influential (late window) satellite orbits, the location of which can have a strong bearing on the impact of observations from different satellites in different regions. Secondly, this sensitivity reinforces the need for data providers to minimize dissemination delays to ensure that crucial late window data reach users in time to be assimilated. Finally, numerical weather prediction (NWP) centres (who run 4D systems) must ensure that these lattermost observations are being captured and used effectively. Some suggestions for this are proposed.