This research aims at developing a highly efficient yet accurate computational framework for the simulation and prediction of wave and wind coupled motions with wave phases being resolved, which will lead to an advanced data assimilation tool to provide more comprehensive environmental input for naval applications. Our ultimate goal is to pave the way for developing an operational tool for the Navy to use for ocean-wave-atmosphere battlespace sensing and prediction with high resolution. The scientific and technical objectives of this research are as follows: (1) use the detailed physics revealed in coupled wind-wave simulations to obtain a fundamental understanding of wave surface-layer processes, based on which physics-based advanced wave-layer models can be developed; (2) adopt a highly accurate immersed boundary method to perform turbulence-wave simulations on a fixed Cartesian grid to achieve superior computation efficiency; and (3) use the developments in (1) and (2) to pave the way for the development of a computational framework for data assimilation with a focus on the reconstruction of wavefield and the retrieval of coherent flow structures from field measurements. This research builds on the combined simulation of wind-wave interactions achieved in a fully dynamical, two-way coupling context. In the simulation, evolution of wavefield is simulated with an efficacious high-order spectral (HOS) method that captures all of the dynamically important nonlinear wave interaction processes. Large-eddy simulation (LES) is performed for the marine atmospheric boundary layer (MABL) in a direct, physical context with wave phases of the broadband wavefield being resolved. In LES, fully resolving the boundary layer at the air-sea interface is prohibitively expensive. We use a wall-layer model to represent the momentum exchange between the flow in the outer layer and the small but dynamically important eddies in the inner layer.