Turbidity current is formed as subaerial open-channel sediment-laden flow plunges into a reservoir. The whole process of reservoir turbidity current, i.e., formation, propagation, and recession, is generally controlled by the water and sediment inputs from upstream and also the reservoir operation scheme specifying the downstream boundary condition. Enhanced understanding of reservoir turbidity current is critical to effective sediment management in alluvial rivers. However, until now there has been a lack of physically based and practically feasible models for resolving the whole process of reservoir turbidity current. This is because the computing cost of three-dimensional modeling is excessively high. Also, single layer-averaged models cannot resolve the formation process characterized by the transition from open-channel sediment-laden flow to subaqueous turbidity current, or the upper clear-water flow as dictated by the operation scheme of the reservoir, which has significant impacts on turbidity current. Here a new two-dimensional double layer-averaged model is proposed to facilitate for the first time whole-process modeling of reservoir turbidity current. The two hyperbolic systems of the governing equations for the two layers are solved separately and synchronously. The model is well balanced because the interlayer interactions are negligible compared with inertia and gravitation, featuring a reasonable balance between the flux gradients and the bed or interface slope source terms and thus applicable to irregular topographies. The model is benchmarked against a spectrum of experimental cases, including turbidity currents attributable to lock-exchange and sustained inflow. It is revealed that an appropriate clear-water outflow is favorable for turbidity current propagation and conducive to improving sediment flushing efficiency. This is significant for optimizing reservoir operation schemes. As applied to turbidity current in the Xiaolangdi Reservoir in the Yellow River, China, the model successfully resolves the whole process from formation to recession. The present work facilitates a viable and promising framework for whole-process modeling of turbidity currents, in support of reservoir sediment management.