The Impact of Dynamic Data Assimilation on the Numerical Simulations of the QE II Cyclone and an Analysis of the Jet Streak Influencing the Precyclogenetic Environment

Abstract

The rapid intensification of a surface cyclone that battered the Queen Elizabeth II (QE II) ocean liner in the western Atlantic Ocean during September 1978 has been the focus of several observational and model-based case studies. The storm is considered a classic example of a cyclone that undergoes explosive deepening, marked by a 60-hPa decrease of the central mean sea level pressure (MSLP) in 24 h. The present study uses a regional-scale numerical model in conjunction with dynamic data assimilation via Newtonian relaxation (or ?nudging?) to provide initial conditions for subsequent simulations of the QE II cyclone. The objectives of this paper are 1) to show that the simulations initialized from the results of 12-h precyclogenetic data-assimilation cycles (with and without bogus data) are superior to those initialized statically from the same data and 2) to resolve the evolution of the upper-level trough-jet system in the 24-h period from 0000 UTC 9 September?0000 UTC 10 September using the dynamically consistent four-dimensional (4D) datasets generated by the model. The 4D model-generated datasets provide the spatial and temporal data resolution not afforded in the observational studies to document the structure and evolution of the dynamical forcing associated with the QE II cyclone. However, the temporal continuity of the cyclone's development, especially the evolution of the upper-level trough-jet system and the associated indirect circulations in the exit region of the upper-tropospheric jet streak, is interrupted at the end of the nudging cycle. This problem poses a limitation for using the 4D datasets for diagnostic studies of the QE II cyclone in the precyclogenetic period during the data-assimilation cycle. Dynamic data assimilation and the inclusion of supplementary data both have a large positive impact on the simulated position and intensity of the QE II cyclone from 1200 UTC 9 September to 0000 UTC 10 September during the initial phase of rapid cyclone development. These runs capture the developing cyclone and associated rate of MSLP falls at 1200 UTC 9 September, whereas the runs based on static initialization delay the deepening six to nine hours into the model simulation. The diagnostic analyses based on these simulations show that the initial development of the QE II storm between 0000 UTC 9 September and 0000 UTC 10 September was embedded within an indirect circulation of an intense 300-hPa jet streak, was related to baroclinic processes that extended throughout a deep portion of the troposphere and was associated with a classic two-layer mass-divergence profile expected for an extratropical cyclone. The runs initialized from data-assimilation cycles, including the bogus data, still underestimate the MSLP of the QE II cyclone by 30% at 24 h into the simulations (1200 UTC 10 September). These results provide further supporting evidence that increasing the horizontal model resolution and improving the subgrid-scale physical parameterizations (especially the precipitation schemes) may be required to simulate the most rapid development phase of the QE II cyclone.