The proper orthogonal decomposition (POD) is among the most trusted tools in the analysis of turbulent fluid mechanics and is frequently employed to identify and isolate dynamically significant features underlying a given flow. The spatially coherent turbulence structures that form the POD modal basis are ordered according to the energy they represent in the full turbulence field, and may be used to filter turbulent flow data or model a flow with only the large-scale dynamics. Herein, the POD is employed as the basis for a reduced-order modeling technique tested on the canonical turbulent channel flow. Afterward, a computationally efficient model for a wind turbine wake (wakeROM) defined through turbulent velocity fields from large-eddy simulation data. The wakeROM is defined by inter-relating dynamic mode coefficients through a series of polynomial parameters and the resulting system of ordinary differential equations models the dynamics of the wind turbine wake using only large-scale turbulence.