Numerical Simulations of a Case of Explosive Marine Cyclogenesis

View More View Less
  • 1 National Center for Atmospheric Research, Boulder, CO 80307
  • | 2 Department of Atmospheric Sciences, University of Illinois, Urbana, IL 61801
© Get Permissions
Full access

Abstract

The extratropical cyclone which damaged the liner Queen Elizabeth II in September 1978 is a well-documented example of explosive marine cyclogenesis in which the 24 h surface central pressure fall was 60 mb commencing 1200 GMT 9 September. Operational models of both the National Meteorological Center (NMC) and Fleet Numerical Weather Central (FNWC) predicted virtually none of the observed surface intensification. This study reports on results of simulations performed with a primitive equation model. Emphasis will be placed on discovering why such poor forecasts were made of this storm. The extensive data set compiled by Gyakum (1983a, b) is used both to initialize and verify the model in a series of 24 h simulations, in order to assess the impact of initializing the model with these supplementary data. Physical processes identified observationally by Gyakum as being important in the storm's evolution are also examined numerically for their relative importance. In a series of seven simulations in which initial condition, horizontal resolution, and physics are varied, the model storm intensity varies considerably. In the weakest, the minimum pressure and maximum boundary-layer wind speeds are 1001 mb and 15.0 m s−1; in the strongest, these parameters are 960 mb and 50.2 m s−1.

The model simulations without the supplementary data set show little improvement over the forecasts of NMC and FNWC. Those simulations with the supplementary data produce improvements in the S1 score, the intensity of the storm and the track of the storm. The improvement in the model simulations with the introduction of the supplementary data appears due to their more realistic documentation of the shallow cyclonic circulation, the small low-level static stability, and enhanced lower-tropospheric water vapor content.

Physical processes also played a major role in the simulators. The effect of surface fluxes of sensible and latent heat were moderate on the 24 h pressure and wind forecasts. In addition, these fluxes produced large changes in the temperature and moisture structure of the planetary boundary layer over a large area of cold northerly flow to the rear of the cyclone.

Latent heating was important in determining the storm intensity and track. Including latent heating through a cumulus parameterization scheme with a horizontal resolution of 90 km produced an improvement in the simulated intensity and position, with a reduction in minimum pressure of 7 mb and an increase in boundary-layer wind speed of 5 m s−1. With 45 km horizontal resolution, use of explicit condensation heating rather than the cumulus parameterization produced a further reduction in minimum pressure of 12 mb. Although experiments with explicit rather than parameterized latent heating produced more intense storms, in agreement with observations, the model storm motion was slowed considerably during the last 6 h of the simulation, resulting in an increased position error.

The model storm showed a small increase in intensity when the horizontal grid length reduced from 90 km to 45 km, with the minimum pressure decreasing by 3 mb. A further reduction in horizontal resolution to 22.5 km produced only minor differences in storm intensity.

The most intense model storm was simulated when an explicit medium-resolution planetary boundary-layer formulation replaced the bulk formulation used in most of the experiments. With 45 km resolution, explicit latent heating, and the medium-resolution boundary-layer model, a storm with minimum pressure of 960 mb and a maximum wind speed of 50.2 m s−1 was obtained.

This study suggests that baroclinic instability in the weakly stratified lower troposphere is the major mechanism of growth for this cyclone, as discussed by Reed, although latent heat plays an important role in the later stages of development. The development of this strong, yet relatively shallow, storm has three major implications for improving operational forecasts Of similar storms. First, the vertical resolution of the model must be adequate; our estimate is that at least four model layers are required below 700 mb. Second, the lower-tropospheric winds, static stability, water vapor content, and sea-surface temperature must be resolved accurately in the initial analysis because of the sensitivity of the model storm to these fields. Third, continued improvement of modeling planetary boundary-layer and latent heating processes is likely to be important in cases of this type.

Abstract

The extratropical cyclone which damaged the liner Queen Elizabeth II in September 1978 is a well-documented example of explosive marine cyclogenesis in which the 24 h surface central pressure fall was 60 mb commencing 1200 GMT 9 September. Operational models of both the National Meteorological Center (NMC) and Fleet Numerical Weather Central (FNWC) predicted virtually none of the observed surface intensification. This study reports on results of simulations performed with a primitive equation model. Emphasis will be placed on discovering why such poor forecasts were made of this storm. The extensive data set compiled by Gyakum (1983a, b) is used both to initialize and verify the model in a series of 24 h simulations, in order to assess the impact of initializing the model with these supplementary data. Physical processes identified observationally by Gyakum as being important in the storm's evolution are also examined numerically for their relative importance. In a series of seven simulations in which initial condition, horizontal resolution, and physics are varied, the model storm intensity varies considerably. In the weakest, the minimum pressure and maximum boundary-layer wind speeds are 1001 mb and 15.0 m s−1; in the strongest, these parameters are 960 mb and 50.2 m s−1.

The model simulations without the supplementary data set show little improvement over the forecasts of NMC and FNWC. Those simulations with the supplementary data produce improvements in the S1 score, the intensity of the storm and the track of the storm. The improvement in the model simulations with the introduction of the supplementary data appears due to their more realistic documentation of the shallow cyclonic circulation, the small low-level static stability, and enhanced lower-tropospheric water vapor content.

Physical processes also played a major role in the simulators. The effect of surface fluxes of sensible and latent heat were moderate on the 24 h pressure and wind forecasts. In addition, these fluxes produced large changes in the temperature and moisture structure of the planetary boundary layer over a large area of cold northerly flow to the rear of the cyclone.

Latent heating was important in determining the storm intensity and track. Including latent heating through a cumulus parameterization scheme with a horizontal resolution of 90 km produced an improvement in the simulated intensity and position, with a reduction in minimum pressure of 7 mb and an increase in boundary-layer wind speed of 5 m s−1. With 45 km horizontal resolution, use of explicit condensation heating rather than the cumulus parameterization produced a further reduction in minimum pressure of 12 mb. Although experiments with explicit rather than parameterized latent heating produced more intense storms, in agreement with observations, the model storm motion was slowed considerably during the last 6 h of the simulation, resulting in an increased position error.

The model storm showed a small increase in intensity when the horizontal grid length reduced from 90 km to 45 km, with the minimum pressure decreasing by 3 mb. A further reduction in horizontal resolution to 22.5 km produced only minor differences in storm intensity.

The most intense model storm was simulated when an explicit medium-resolution planetary boundary-layer formulation replaced the bulk formulation used in most of the experiments. With 45 km resolution, explicit latent heating, and the medium-resolution boundary-layer model, a storm with minimum pressure of 960 mb and a maximum wind speed of 50.2 m s−1 was obtained.

This study suggests that baroclinic instability in the weakly stratified lower troposphere is the major mechanism of growth for this cyclone, as discussed by Reed, although latent heat plays an important role in the later stages of development. The development of this strong, yet relatively shallow, storm has three major implications for improving operational forecasts Of similar storms. First, the vertical resolution of the model must be adequate; our estimate is that at least four model layers are required below 700 mb. Second, the lower-tropospheric winds, static stability, water vapor content, and sea-surface temperature must be resolved accurately in the initial analysis because of the sensitivity of the model storm to these fields. Third, continued improvement of modeling planetary boundary-layer and latent heating processes is likely to be important in cases of this type.

Save