Modification and Application of the Navy Global Environmental Model (NAVGEM) to Generate High-Altitude (0-100 km) Global Atmospheric Reanalyses of the 2014 Austral Winter for the Deep Propagating Gravity Wave Experiment (DEEPWAVE)

The Deep Propagating Gravity Wave Experiment (DEEPWAVE) was an international aircraft-based field program to observe and study the end-to-end dynamics of atmospheric gravity waves over a deep altitude range (0-100 km) and the effects of these waves on atmospheric circulations. DEEPWAVE took place from an operating base in Christchurch, New Zealand during May-July of 2014. Intense orographic and nonorographic gravity waves were observed from the lower troposphere to deep into the mesosphere and lower thermosphere (MLT). Understanding the dynamics of these deep wave events requires knowledge of background winds and temperatures from 0-100 km, yet these aircraft observations occurred within the remote greater New Zealand and Southern Ocean regions where few if any ground-based wind and temperature observations were available. Furthermore, operational data assimilation systems (DASs) provide wind and temperature information up to ~50-70 km altitude only, leaving a critical MLT analysis gap from ~60-100 km that inhibits DEEPWAVE science.

To address this gap we developed a high-altitude (0-100 km) configuration of the Navy’s global numerical weather prediction system (NAVGEM) and used it to generate 0-100 km reanalyses of the 2014 austral winter for DEEPWAVE science. We discuss aspects of this system, including new physical parameterizations and assimilated observations for the 50-100 km altitude range. We compare 4 separate reanalyses generated using: (a) two separate forecast model configurations, one with low horizontal resolution (T119) and another with high (gravity-wave-resolving) horizontal resolution (T425), and; (b) two different DA algorithms, one based on a pure four-dimensional variational (4DVAR) algorithm with static background error covariances, and another based on a new hybrid-4DVAR algorithm where 80-member forecast ensembles modify background error covariances at all altitudes.

Acknowledgements: The NRL components of this research were supported by the Chief of Naval Research and by the Office of Naval Research.