The prediction of tropical cyclone (TC) motion has improved greatly in recent decades. However, similar trends remain absent with respect to TC intensity prediction. Several hypotheses have been proposed attempting to explain why dynamical NWP models struggle to predict TC intensity. The leading candidates are as follows: (1) the lack of an evolving ocean (i.e., sea-surface temperature) boundary condition which responds as a function of the atmosphere (e.g., TC) forcing, (2) inappropriate initial conditions for the TC vortex (e.g., lack of data assimilation methods), (3) NWP model grid-length resolutions which are unable to resolve the temporal and length scale for the features believed responsible for TC vortex intensity. modulations (i.e., eye-wall dynamics, momentum transport, vortex Rossby wave interactions, etc.), and (4) physical parametrization which do not adequately represent the air-sea interactions observed during TC passage. In this study, a coupling algorithm for two independent, high-resolution, and state-of-the-art atmosphere and ocean models is developed. The atmosphere model -- the Advanced Weather Research and Forecasting (WRF-ARW) model is coupled to the HYbrid Coordinate Ocean Model (HYCOM) using a (UNIX) platform independent and innovative coupling methodology. Further, within the WRF-ARW framework, a dynamic initialization algorithm is developed to specify the TC vortex initial condition while preserving the synoptic-scale environment. Each of the tools developed in this study is implemented for a selected case-study: TC Bertha (2008) and TC Gustav (2008) for the coupled-model and TC vortex initialization, respectively. The experiment results suggest that the successful prediction (with respect to the observations) for both the ocean response and the TC intensity cannot be achieved by simply incorporating (i.e., coupling) an ocean model and/or by improving the initial structure for the TC. Rather the physical parametrization governing the air-sea interactions is suggested as the one of the weaknesses for the NWP model. This hypothesis is (indirectly) supported through a diagnostic evaluation of the synoptic-scale features (e.g., sea-level pressure and the deep-layer mean wind beyond the influence of the TC) while the assimilated TC vortex is nudged toward the observed intensity value. It is found -- in the case of TC Gustav (2008) using WRF-ARW, that as the assimilated TC vortex intensity approaches that of the observed, the balance between the mass and momentum states for WRF-ARW is compromised leading to unrealistic features for the environmental sea-level pressure and deep-layer (800- to 200-hPa) mean wind surrounding the TC. Forcing WRF-ARW to assimilate a TC vortex of the observed maximum wind-speed intensity may ultimately compromise the prediction for the TC's motion and subsequently mitigate any gains for the corresponding intensity prediction.Suggestions for additions to the coupled atmosphere-ocean model include a wave-model (WAVEWATCH3), the assimilation of troposphere thermodynamic observations, and modifications to the existing atmospheric boundary-layer parametrization. The current suite of atmosphere model parametrizations do not accurately simulate the observed azimuthal and radial variations for the exchange coefficients (e.g., drag and enthalpy) that have been indicated as potentialpredictor variables for TC intensity modulation. However, these modifications should be implemented only after the limitations for the current coupled-model and TC vortex initialization methods are fully evaluated.
