Analysis Of a Tropical Cyclone Boundary Layer Database Against MM5 Model Simulations

It is well-known that ocean heat fluxes compensate adiabatic expansional cooling as PBL air flows toward the lower pressure of the tropical cyclone inner-core, thereby maintaining eyewall buoyancy and allowing the storm to intensify under favorable environmental conditions. However, recent oceanic observations of cooling in the hurricane PBL show this compensation may be less than originally believed (Pudov 1992; Black 1995). Cione et al (1998) have recently analyzed Atlantic buoys and verified this cooling, but noted most of it occurred outside the inner core. If such observations represent the majority of tropical cyclones, it would raise several issues about the tropical cyclone intensification process.

In an effort to increase the tropical cyclone PBL sample, a database containing oceanic observations from the Atlantic, western North Pacific, and the Southern Hemisphere has been developed. These platforms include CMAN and moored buoys from the National Data Buoy Center and the Japanese Meteorological Agency for 1974-1996, as well as buoys, ships, and atoll observations from the Southern Hemisphere for 1975-1997. This expanded dataset is being compared against the findings of Cione et al.

Tentative observations support the conclusions of Cione et al.: much of the cooling occurs outside the inner-core, and budget calculations attribute much of this cooling to rainband downdrafts. Ocean mixing and adiabatic expansion are usually not contributors to cooling. Minimal cooling occurs within the eyewall, suggesting that an external heating source is compensating adiabatic expansion. This external heating source is likely sensible heat flux, but we hypothesize some of this heating is due to recycled eye air entrained into the eyewall PBL. Wet-bulb computations also suggest the inner-core cooling is partially due to sea spray evaporation.

To assess the capability of numerical models in simulating these observations, the PBL of an MM5 Hurricane Andrew simulation will be examined. This simulation was performed on a massively parallel IBM SP2 using 16 processors, and is initialized at the moment preceding rapid intensification. The vortex was spun-up using reconnaissance flight data from the Hurricane Research Division. The PBL of this simulation will be compared against buoy observations. Other simulations of Hurricane Hugo and Hurricane Mitch may also be presented.