4D-Var data assimilation in a nested, coastal ocean model: A Hawaiian case study

State estimation techniques have been well established in open ocean systems; however they are less often used in coastal applications due to nonlinearity. Using 4D-variational data assimilation in a triple one-way nested system, we investigate the processes that control coastal dynamics for a test case along the western coast of Oahu, Hawaii. All available observations are combined with the model dynamics for 13 months. Over this time, the residual error between the model and observations was improved by nearly 30% in the surface temperature and 34% in the alongshore ADCP currents. The barotropic and baroclinic tides are found dominate the local circulation; however, island and atmospheric interaction generates an island wake effect that is important to the subtidal dynamics of the region. The baroclinic tides exhibit well-defined energy paths, and the initial condition corrections, despite altering the density waveguide, have little influence on the propagation of the baroclinic energy, which is controlled by the propagation of baroclinic tides generated outside of the domain. We find the larger-scale, advected dynamics control the local surface circulation through boundary condition adjustment, accounting for 45% of the total corrections made via data assimilation system. The initial conditions controls little of the evolution of this local, coastal flow and has a short persistence. The wind stress control vector is important in the central region of the domain inducing flow toward the lee of the island. Our results reveal that coastal studies may not be initial value problems, rather they are forced problems that require a knowledge of the large-scale energy propagated into the region.