Lisa Neef, Ted Shepherd (University of Toronto) Saroja Polavarapu (Meterological Service of Canada)
In recent years, much research has focused on the control of spurious =20 inertia-gravity waves within 4D data asssimilation. Most of this =20 work has focused on avoiding the generation of spurious gravity waves =20 in cases where slow vortical motion is of primary interest, and =20 gravity waves are considered noise (e.g. Neef et al. 2006). However, =20 there are many applications where the true state contains =20 nonnegligible gravity waves, such as the mesosphere. The advent of =20 4D data assimilation, and the growing inventory of available =20 observations, makes capturing gravity waves within an assimilation a =20 possibility -- with many caveats.
Our work examines the problems and issues inherent in such problems, =20 using low order models which admit motion of multiple timescales, but =20 are yet simple enough to make interpretation of experiments =20 straightforward. This talk will in particular show examples of how =20 and when standard 4DDA algorithms succeed (or fail) at capturing =20 states where the ``primary'' motion is contaminated by comparatively =20 fast waves. We will also offer an interpretation of recently-=20 proposed modifications to the standard algorithms (such as ensemble =20 inflation or square root filtering), in terms of the gravity wave =20 problem described above.