Comments on ‘‘A Multiscale Numerical Study of Hurricane Andrew (1992). Part I: Explicit Simulation and Verification’’

In their paper ‘‘A multiscale numerical study of Hurricane Andrew (1992). Part I: Explicit simulation and verification,’’ Liu et al. (1997, hereafter referred to as LZY) present an impressive simulation of a mature tropical cyclone initialized with environmental data and operational model analyses obtained during Hurricane Andrew. The simulated tropical cyclone makes landfall near Palm Beach, Florida, with an eye diameter about twice that observed. The minimum central sea level pressure at landfall is similar to that observed when Hurricane Andrew made landfall about 105 km farther south near Homestead, Florida. According to Fig. 2 of LZY, the simulated storm begins filling at landfall and its pressure increases to about 944 hPa in 5 h, similar to that observed in Andrew (Powell et al. 1996; Powell and Houston 1996, hereafter referred to as PHR and PH). LZY compared the surface wind field of the simulated storm to that observed in Hurricane Andrew by PH and suggested that the coastal discontinuities published in PH were the product of reducing flight-level winds to the surface with a planetary boundary layer (PBL) model, analyzing land and marine observations separately, and then merging them at the coastline ‘‘without including full dynamic and physics interactions.’’ LZY provided little information on the prelandfall wind distribution and land surface properties to substantiate the landfall wind field produced by their model. The reply to this comment (Zhang et al. 1999, hereafter referred to as ZLY) indicates that the landfall winds described in LZY correspond to a completely different framework for exposure, height, and averaging time than that used by PH. The lack of coastal discontinuities in LZY’s landfall wind field corresponds to a distinct prelandfall wind distribution that differs both from the observations and earlier hours of the simulation.