Adjoint-derived forecast senstivity gradients have been used for a variety of purposes including designing targeted observing strategies and identifying 'key' analysis errors. To date, few studies have explored the relationship of these sensitivity gradients to characteristics of the ambient flow about which they are calculated, nor have many studies provided consistent synoptic and dynamical interpretations of the sensitivity gradients or their evolution.

In this presentation, the results of an on-going research effort at the Univ. of Wisconsin-Madison to provide meaningful synoptic interpretations of adjoint sensitivity fields calculated in real-time will be provided. The response functions chosen for this study include an 'energy-weighted' forecast error, area-averaged temperature, and the circulation about a geographical region at a prescribed forecast interval. The corresponding sensitivity gradients are evaluated with respect to the model state vector as well as with respect to derived quantities including the relative vorticity and potential vorticity.