ABSTRACT: A numerical model coupled with data-driven statistical method is proposed to forecast real-time significant wave heights for a ship route during hurricanes. The proposed model can be used for determining the wave heights of buoys and the wave heights on the ship trajectory, considering the short-term time step (15 min) of the ship operation. A shipping line to the Caribbean Sea and the Gulf of Mexican was used to simulation purpose. We used artificial neural network-based multi-layer perceptron to develop the wave height prediction model. The statistic quadtree-adaptive model was used to generate the input-output relationships. To compare the efficiency achieved using ANNs, traditional multiple linear regression was selected as a benchmark. Results showed ANN-based prediction models can be regarded as being reliable, and the proposed wave height forecast model can be effectively used in forecasting.