Sensitivity of Model Simulations for a Coastal Cyclone

Abstract

A nested global, limited-area model was used to predict the President's Day cyclone of 18-19 February 1979. Both a low (?150 km) and a high (?50 km) horizontal resolution version were used. The model has full physics with a planetary boundary layer, moisture, moist convective adjustment, and radiation. The low-resolution model using a global analysis for initial and boundary conditions (termed a simulation), was able to capture the general development and movement of the cyclone. Some discrepancies were noted for the intensity of upper-air features between the analyses and the model solution during the first 24 hours. The primary focus of this paper is to determine the effect of initial and boundary conditions, as well as model parameterizations on the accuracy of the predictions. The evolution of the storm is discussed with an emphasis on the quality of the numerical simulation. The impact of the initial conditions on the model solution was tested by using four different global analyses. It was found that the variability between the solutions was less than the variability between the analyses. Varying the horizontal diffusion in the model produced stronger development with weaker diffusion, but the character of the development did not change significantly. The sensitivity of the simulation to latent heat was tested by running the model without latent heating. A low did develop in this model solution, although it was much weaker and it did not develop vertically as in the cases with latent heating. The most significant improvement in accuracy in this sensitivity study occurred when the horizontal resolution was increased from 1.25° ? 1.0° (?150 km) to 0.4° ? 0.32° (?50 km). The position and intensity of the surface low were much close to reality, as indicated by comparison with a mesoanalysis and to satellite pictures. The nested model was also run in forecast mode with boundary conditions for the limited-area model supplied by the (Geophysical Fluid Dynamics Laboratory) GFDL global model forecast. In general, the quality of the limited-area forecast compared very well with the simulations. The overall character and intensity of the development were similar. The role of lateral boundary conditions was demonstrated by comparing forecasts and simulations with identical initial conditions. The results suggest the increasing importance of the boundary data with time in the limited-area forecast and show high correlation between the errors in the limited-area forecast and the global forecast within the limited-area domain.