Wednesday, 1 August 2001
Balanced initialization procedures and mesoscale predictability
The danger of imposing too strong of a balance constraint in the initialization of mesoscale models, such as through 3D variational analysis and nonlinear normal mode initialization methods, is demonstrated through detailed diagnostic analyses of a numerical simulation of a well-observed, large-amplitude gravity wave event that produced hazardous winter weather. Wavelet analysis is used to investigate the evolving gravity wave structure in the simulation and potential vorticity (PV) inversion is employed to study the nature of the flow imbalance in the wave generation region. The residual in the nonlinear balance equation and the unbalanced geopotential height field obtained from PV inversion are evaluated for their usefulness in diagnosing the flow imbalance. Both of these fields showed clear evidence of strong imbalance associated with a middle-to-upper tropospheric jet streak and tropopause fold upstream of the large-amplitude gravity wave several hours before the wave became apparent at the surface. Although this dynamic imbalance existed in the initial model state, and rapidly grew in intensity during the first 4h of the simulation, gravity waves resulting from the initial imbalances had fully subsided by 2 hours into the model forecast, allowing the subsequent growth of physically meaningful gravity waves to develop by nonlinear processes in a manner consistent with the observations. The numerical model was run without four-dimensional dynamic assimilation or any special initialization method; in fact, a run with 12 hour pre-forecast nudging produced an inferior simulation of the gravity waves. Furthermore, the results from various model sensitivity simulations indicate that diabatic heating played an important role in the mesoscale dynamics. A weaker upper-level jet, cyclogenesis, and gravity wave activity all occurred in a simulation in which diabatic processes related to the phase change of water substance were not permitted at all or prevented for a fixed period of time. These results suggest that the use of strong balance constraints and procedures for initializing mesoscale models is not advisable. Gravity waves were continuously generated at the tropopause by geostrophic adjustment in the exit region of the unbalanced upper-level jet streak as it approached the inflection axis in the height field immediately downstream of the maximum imbalance associated with the tropopause fold. A split front in the middle troposphere, characterized by the advance of the dry conveyor belt above the warm front, was overtaken by one of these propagating waves. During this merger process, a resonant interaction resulted, which promoted the rapid amplification and scale contraction of both the incipient wave (nonlinear wave development) and the split front (frontogenesis). The strong gravity wave and front aloft became inseparable following this merger. These dynamical processes and the ensuing development of deep convection within this front-wave merger zone in the model simulation are highly realistic, again lending support to the idea that allowing for a degree of mass-momentum imbalance in the model’s initial dynamical balance is appropriate at the mesoscale.