Impact of Initial Conditions, Rainfall Assimilation, and Cumulus Parameterization on Simulations of Hurricane Florence (1988)

Abstract

Numerical simulations were performed with the Pennsylvania State University/National Center for Atmospheric Research Mesoscale Model Version 5 (MM5) to study the impact of initial conditions, satellite-derived rain assimilation, and cumulus parameterization on Hurricane Florence (1988). A few modifications were made to the J. Manobianco et al. (MKKN) rain assimilation scheme, which was developed originally for midlatitude weather systems, to successfully simulate organized tropical weather systems such as Florence. These changes consist of replacing latent heating scaling with convective rainfall in the Kuo?Anthes scheme in areas where both the model-predicted and satellite-derived rainfall coincide, and specifying a normalized parabolic heating profile in deep convective regions where there is satellite rain but no model rain. Restoration of the original Kuo?Anthes heating distribution function in lieu of the fixed heating profile specified in the MM5 model is another change implemented in the Kuo?Anthes scheme. Results from the sensitivity simulations made with the modified rain assimilation scheme show that 1) the enhanced initial conditions with the omega dropsonde data yield a positive impact on the development of Florence for both the Betts?Miller and the modified Kuo?Anthes schemes, 2) the effect of ingesting continuous (Special Sensor Microwave/Imager and Geostationary Operational Environmental Satellite Infrared) satellite-derived rainfall rates as latent heating by the modified rain assimilation scheme is much greater with the modified Kuo?Anthes scheme than with the Betts?Miller scheme, and 3) the combined impact of enhanced initial conditions and rain assimilation yields a superior simulation of Florence, particularly with the Kuo?Anthes scheme. The weak response of the Betts?Miller scheme to rain assimilation, compared to the large impact with the Kuo?Anthes scheme, appears to be related mainly to the differences in the treatment of convective rainfall and its latent heat release in respective cumulus parameterization schemes. Since the MKKN scheme mainly invokes latent heat scaling to ingest satellite rainfall, the Kuo?Anthes scheme responds to increased latent heating from satellite rainfall rates more favorably through conditional instability of the second kind (CISK)-type feedback effects than the Betts?Miller scheme. The latter result clearly suggests that the success of the modified rain assimilation scheme on development of organized tropical systems such as Hurricane Florence depends to a large extent on the choice of cumulus parameterization scheme.