Recent workshops and papers have highlighted the shortcomings of existing reanalyses (e.g., NCAR/NCEP, ERA-40) for application to the Arctic and Antarctic. As a consequence, a preliminary effort is underway to attempt to design a regional reanalysis system that would be more appropriate for high latitude application. A requisite key component of such a regional reanalysis system will be a four-dimensional data assimilation approach that: a) adequately ingests all available data sources in the Arctic, including polar-orbiting satellite data depicting atmospheric, ice and ocean states; b) adequately takes into account the differing structural elements of the polar atmosphere (e.g., boundary-layer and surface-based inversions, relatively low liquid and total water content, and strong variability in surface conditions on the mesoscale, particularly over sea ice; c) utilizes a state-of-the-art mesoscale modeling system as the dynamic integration mechanism.

Developing such a reanalysis system is a daunting undertaking requiring significant preparatory steps, which are presently underway at the University of Alaska Fairbanks, Ohio State University, the University of Colorado and the University of North Dakota. In this paper, we describe our initial efforts and tests of a mesoscale three-dimensional variational assimilation system, the MM5-3DVAR system developed at the National Center for Atmospheric Research, to high latitudes. This system was selected due to its commonality with the system, also developed at NCAR, available with the emerging state-of-the-art, community-developed Weather Research and Forecast Model, presently being modified for Arctic use by Ohio State University and NCAR, a work that is presented elsewhere in this conference.

In our poster presentation, we describe the methodology we have used to implement MM5-3DVAR for the Arctic, as well as the experiment design used in tests of the system for case studies covering the entire annual cycle as well as specific extreme event cases to be described in detail in a companion paper. Our tests include examination of the sensitivity to horizontal resolution utilized in the 3DVAR analysis as well as 3DVAR methodology (static versus cycling), as well as the efficacy of the current MM5-3DVAR formulation for conventional observations as opposed to other assimilation methodologies such as Newtonian nudging, available as a standard feature within the MM5 modeling system.

We will present results from a selection of the case studies examined thus far on both a pan-Arctic domain and a Western Arctic domain centered on Alaska. The methodology behind and results from efforts to incorporate retrieved satellite quantities from polar-orbiting platforms (such as MODIS) will be discussed briefly, with a more detailed description presented elsewhere in the conference.