Nicholas Kedzuf, Shuyi S. Chen, and Brandon Kerns
University of Miami
Rainfall associated with tropical cyclones (TCs) at landfall is a leading cause of casualties and property damage throughout the world. The factors and physical processes that influence TC precipitation are complex and not well understood. As a result of this, the skill of TC quantitative precipitation forecasts (QPFs) in numerical models is still relatively poor. The main objectives of this study are to 1) better understand how the most fundamental TC characteristics, such as its propagation speed, intensity, and storm size affect the distribution, duration, and intensity of TC rainfall, and 2) evaluate TC QPF in a cloud-resolving coupled atmosphere-ocean model.
In this study, model forecasts of five TCs of various intensity and track in the Atlantic and Pacific basins, including Hurricanes Ike (2008), Earl (2010), Isaac (2012), and Sandy (2012) and Typhoon Fanapi (2010) are analyzed against observations. We use observational data from both the Tropical Rain Measuring Mission (TRMM) satellite and National Hurricane Center (NHC) Best Track estimates. The model used in this study is the Unified Wave INterface-Coupled Model (UWIN-CM) with nested model grids of 12, 4, and 1.3 km resolutions. To investigate the effects of air-sea coupling on model QPF in TCs, three numerical experiments for each TC are included: an uncoupled atmosphere (UA) control, an atmosphere-ocean (AO) coupled, and an atmosphere-wave-ocean (AWO) coupled configuration.
The QPFs are presented in three variables: storm-averaged rain rate (mm/h), daily storm total accumulation at each grid point (mm/day), and volumetric rainfall (m3). The first two may be more useful for TC impact forecasts, while the last is a good measure for evaluating a storm’s water budget. In order to investigate the relationship between TC rainfall and other properties, we first compare TC characteristics from Best Track data with the model results. As the high-frequency change in storm intensity and rainfall can be noisy, we use daily averages and storm totals to provide a consistent temporal platform throughout our analysis.
When UWIN-CM output was compared against observations across all storms, we found that all three model configurations had a strong tendency to overestimate QPF variables. As expected, the UA configuration performed the poorest; overestimating average rain rates by 35% and total volumetric rainfall by 36% (of the order of billions of cubic meters of rainfall). The configurations with higher degrees of coupling (AO/AWO) were more accurate and quite similar to each other, overestimating average rain rates and volumetric rainfall by 10-12%. We found that inner-core and near-core azimuthal rain rate averages were 3-4x higher in UWIN-CM storms. While each configuration exhibited similar rain rate structures; the UA configuration produced the highest near-core rain rate maxima followed by the AO and AWO configurations. All configurations performed well from ~100-150 km to the edge of the storm. Model storms also had consistently smaller fractional areas of rain rate > .1 mm/hr than TRMM regardless of radial distance. This could potentially be the result of issues with the accuracy of TRMM in resolving near-core precipitation as well as issues with model storm evolution.
![](images/flag.png)
![](images/blue_ribbon.png)