Effects of Surface Roughness Representation on Hurricane Intensity and Structure Forecasts

Wednesday, 20 April 2016: 11:00 AM
Ponce de Leon C (The Condado Hilton Plaza)
Yi Jin, NRL, Monterey, CA; and S. W. Wang, J. R. Moskaitis, and J. Doyle

A crucial issue in the numerical prediction of tropical cyclones is the parameterization of momentum, heat and moisture fluxes across the ocean-atmosphere interface. The magnitude of these fluxes depends largely on how momentum and heat/moisture transfer coefficients of the surface flux parameterization are specified. In particular the surface drag coefficient, CD, representing the efficiency of the atmospheric kinetic energy dissipation into the underlying ocean, has been a focus in numerous observational and numerical modeling studies for high-wind conditions. It has been suggested in previous studies that CD levels off and even decreases with increasing winds when the surface wind speed exceeds 20-35 m s-1. These new findings have been incorporated and evaluated in the Coupled Ocean/Atmosphere Mesoscale Prediction System for Tropical Cyclones (COAMPS-TC) with the purpose of improving TC intensity forecasts. In COAMPS-TC the drag coefficient CD and heat/moisture coefficients (CH/CQ) are functions of surface roughness lengths for momentum (z0) heat (z0h) and moisture (z0h), based on the similarity theory. In this study, COAMPS-TC is used as a test bed to examine the impact of various z0 calculations derived from several different parameterizations on TC simulations. Our sensitivity tests suggest that predictions of TC intensification and structure are highly sensitive to the specification of the roughness lengths. The optimal z0 formulation results in a reduction in a negative intensity forecast bias, especially for cases that reached hurricane intensity at the model initialization time. The pressure-wind relationship from simulations using the new formulation for z0 has improved considerably for the high-intensity range. Rapid intensification, as well as Cat-4 and Cat-5 storms are attainable and in agreement with observed estimates, so that the relative frequency of storm distribution with respective to intensity is more realistic using the new z0.
- Indicates paper has been withdrawn from meeting
- Submission entered in competition