Ensemble Adaptive Data Assimilation Techniques Applied to Land-Falling North American Cyclones - Data Assimilation for Atmospheric, Oceanic and Hydrologic Applications - page 556

Fig. 23.4 Level of maximum variance reduction of the response function for all cyclones for
observations of ( a ) temperature, ( b ) pressure, ( c ) zonal wind, and ( d ) meridional wind

The fact that targeting regions are clearly co-located with features in the flow is
likely a unique feature of ensemble targeting techniques. Following the discussion
above, the impacts from hypothetical observations within an EnKF highlight
specific flow features through their relationship to dynamically sensitive areas. In
turn, these relationships depend on how the specific features in each ensemble
member covary with the dynamically sensitive regions, and are thus strongly
linked to the flow dependence present in the atmospheric state at any given time.
Consequently, covariances that do not possess such flow dependence are unlikely
to capture these relationships, and targeting regions based on 3DVAR systems will
probably differ to some degree from those based on ensemble methods. This further
stresses the importance of directly accounting for the specific assimilation system
when calculating the impacts of targeted observations.

23.3.4 Characteristics of Observation Targeting

for all 2009/2010 Cyclones

Figure 23.4 represents the level where the maximum variance reduction exists for
each cyclone for pressure, temperature, and wind observations. For both temperature