Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Civil and Environmental Engineering


Dr. Craig Miller


As the medium between the sea surface and the upper atmosphere in a hurricane, the Hurricane Boundary Layer (HBL) plays a key role in the overall dynamics of a tropical cyclone, and therefore turbulence exchanges within the HBL deserve a thorough investi- gation. However, since it is dangerous and difficult to take direct measurements within the HBL, studies of the HBL turbulence processes based on direct observations are rare. Thanks to the newly developed dropwindsonde equipped with a Global Position System (GPS) receiver, it is now possible to measure wind velocities and other meteorological variables with an unprecedented accuracy and resolution in the HBL.

To fully utilize dropwindsonde measurements, it is necessary to thoroughly understand its motion characteristics in the measured wind field since its horizontal motions are usually reported as wind measurements. For that reason, the dropwindsonde motion in a pseudo-stochastic wind field with known statistics is simulated. The simulation results illustrate the importance of the wind finding equations introduced by Hock and Franklin (1999) which calculate local winds from dropwindsonde motions. The simulation results show that they are important in reproducing both mean and turbulent wind structures in the HBL. One of its basic assumptions that the dropwindsonde drag coefficient is a constant regardless of the angle of attack is, however, invalidated by the wind tunnel tests conducted in this study. Given that this assumption is essential in both deriving the wind finding equations and in conducting the numerical dropwindsonde motion simulation described above, it is necessary to adapt the dropwindsonde motion model and to repeat the motion simulation to recheck the validity of the wind finding equations demonstrated in the previous simulation. The results validates the wind finding equations although it is derived based on a false assumption that the dropwindsonde drag coefficient is a constant regardless of angles of attack. Moreover, through analyzing the adapted dropwindsonde motion model, a new approach to estimate the vertical wind is proposed which is shown to increase the accuracy in vertical wind estimation by nearly 70%.

Based on the findings derived in the dropwindsonde motion simulations, an in-house software package is designed to process the actual dropwindsonde measurements gathered from 1997 to 2010. The in-house software package, showing an effectiveness equivalent to other widely used processing systems, gives users more control over the processing and compositing procedures used to derive the desired statistics of the measured wind field. With the help of this software package, dropwindsonde measurements are processed and composited to produce the mean, turbulence intensity, and turbulent length scale profiles of the HBL. While the mean wind structure confirms the findings made by several previous studies, the turbulence structure reveals that the turbulence diffusivity formulation currently used by the Yonsei University planetary boundary layer scheme, or the YSU scheme, in Weather Research and Forecasting Model(WRF), a widely used hurricane wind simulation package, correctly simulate turbulent mixing in the HBL up to 200m from the sea surface. In a theoretical discussion of the validity of the YSU scheme, it is found that both the velocity scale and height scale used in its turbulence diffusivity formulation should be revised to take into consideration the special turbulence characteristics in the HBL.

For the purpose of checking the turbulence diffusivity formulation used in the YSU scheme, high resolution numerical simulations of an idealized tropical cyclone are con- ducted using WRF. The simulation results show that only revising the HBL height calculation is not adequate to improve the numerical simulation of hurricanes. Therefore, a deeper investigation of the YSU scheme in simulating the HBL turbulence is required.