Numerical simulations of the genesis of Hurricane Diana (1984). Part II: Sensitivity of track and intensity prediction

The authors examine numerous simulations that probe the dynamics governing the intensification and track of Tropical Cyclone Diana (1984) simulated in Part 1. The development process is fundamentally dependent on a preexisting upper-tropospheric trough-ridge couplet. This couplet focuses mesoscale vertical motion that, in turn, produces lower-tropospheric potential vorticity (PV) anomalies, which form the seed for the tropical storm. Removal of this trough-ridge couplet from the initial conditions eliminates cyclogenesis. The simulated rate of development of Diana in the prehurricane phase depends principally on choices of cumulus parameterization, boundary layer treatment, sea surface temperature, and grid spacing. Simulations with cumulus schemes that allow more grid-scale precipitation on the 9-km grid exhibit unrealistic grid-scale over-turning and slower intensification, primarily due to production of cyclonic vorticity anomalies at large radii. Use of an innermost nest with 3-km grid spacing, without a cumulus scheme, generally produces the intensification that agreed best with observations. Improvement over the control simulation stems from the emergence of convective downdrafts and a vertical motion spectrum that is less biased toward ascent. Consistent with recent work by Braun and Tao, the medium-range forecast model (MRF) planetary boundary layer (PBL) scheme produces a PBL that is too dry too deep as winds intensify toward hurricane strength. Use of the Burk-Thompson scheme leads to excessive intensification with a 9-km grid spacing. Manual analysis of surface data produces a sea surface temperature (SST) field 1°-2°C warmer than is obtained from operational analysis. The warmer SSTs result in a storm that is about 27 hPa deeper after 60 h of integration. Storm track depends primarily on synoptic-scale structure at upper levels. Cumulus schemes that allow more grid-scale overturning enhance the anticyclonic outflow aloft. The outflow deforms the tropopause, building an anticyclone poleward of the storm and facilitating cutoff low formation equatorward of the storm. Using PV attribution, it is shown that these upper-level changes are responsible for an enhanced easterly steering flow and more westward storm track. Later initialization allows a better analysis of trough fracture, particularly the cutoff low and this also leads to a more westward storm track. Overall, despite the presence of a well-defined baroclinic precursor, the large sensitivity of track and intensity prediction to variations in model physics and initial conditions suggests that the development of Diana pushes the current limits of predictability.