In recent years, ocean development has been accelerated, and it has been required to improve the efficiency and safety of marine operations. If the prediction of individual waves in real time is possible, it is expected to increase the operation rate and its safety. we proposed a hybrid prediction procedure in which the physical model and the neural network model are combined. Virtual intermediate points are defined as fixed points near drones. The wave elevation at intermediate points is predicted using the physical model with a measured data by drones. The displacement of the prediction point is predicted by the neural network from the wave elevation at intermediate points predicted by the physical model.