The diurnal transition of the atmospheric boundary layer from thermally stable to unstable stratification subjects wind turbines to rapidly evolving inflow, yet the wake dynamics during this non-equilibrium period remain poorly characterized. Here we use measurement-assimilated large-eddy simulation to reconstruct the boundary-layer evolution observed on 26 September 2023 at the King Plains wind farm during the American WAKE experimeNt (AWAKEN) field campaign, and we analyze the resulting turbine wake and power production across three regimes identified during the transition: stable, transitional, and unstable. As the boundary layer becomes increasing unstable, the hub-height turbulence intensity increases by roughly an order of magnitude compared to stable night-time conditions and the characteristic length scale of atmospheric motion grows from a few rotor diameters to the full boundary-layer depth. Resolvent analysis, applied around the time-averaged wake in each regime, show that the dominant wake-turbulence mechanism shifts from a single low-rank, shear-driven coherent structure under stable stratification to a higher-rank, buoyancy-driven state under unstable stratification, with the transitional regime exhibiting hybrid behavior. The array-averaged power, its variability, and its spectral content change significantly over a period of a few hours, demonstrating that steady-state stability classifications do not adequately represent the dynamics of the diurnal cycle.