Search for: All records

This is the Masters thesis of Badrul Hasan from UMBC in the Department of Mechanical Engineering. The co-main advisors are Meilin Yu (UMBC) and Stephen Guimond (UMBC, now with Hampton University). Numerous aspects of human existence, both material and immaterial, can be disrupted by a hurricane. In this work, the computational fluid dynamics of hurricane rapid intensification (RI) are studied by running idealized simulations with two different codes: a community-based, finite-difference/split-explicit model (WRF) and a spectral-element/semi-implicit model (NUMA). Rapid intensification is what RI stands for, how a hurricane gets stronger quickly. The main goal of this study is to find out how implicit numerical dissipation (IND) affects the energy of the vortex's response to heating, which describes the fundamental dynamics of the hurricane RI process. The heating that is taken into account here is derived from data. These observations include four-dimensional, fully nonlinear, latent heating/cooling rates estimated using airborne Doppler radar readings acquired during RI in a hurricane. The results show that WRF has more IND than NUMA, with a decrease in several intensity parameters, such as (1) the time-integrated mean kinetic energy values that WRF predicts are 20% lower than those that NUMA predicts and (2) the peak, localized wind speeds that WRF predicts are 12 meters per second slower than those that NUMA predicts. To make a time series of intensity similar to NUMA, the eddy diffusivity values in WRF need to be less than those in NUMA by about 50%. Various analyses are conducted comparing WRF with NUMA. Kinetic energy budgets reveal that the pressure contribution is the primary cause in the model variations, with WRF generating an average 23% lower vortex energy input. The IND is associated with the low-order spatial discretization of the pressure gradient in WRF. In addition, the mean contribution of the eddy transport term to the vortex intensification is determined to be 20% positive. These findings have significant implications for the academic and operational forecasting communities that employ WRF and similar numerical methods.