Wave height forecasting is very important for coastal management and offshore operations. In this paper, the accuracy and performance of three soft computing techniques [i.e., Multi-Layer Perceptron (MLP), Radial Basis Function Neural Network (RBFNN) and Adaptive Neuro Fuzzy Inference System (ANFIS)] were assessed for predicting significant wave height. Using different combinations of parameters, the prediction was done over a few or a two days’ time steps from measured buoy variables in the Caspian Sea (case study: Anzali). The data collection period was from 03.01.2017 to 06.01.2017 with 30-minute intervals. The performance of different models was evaluated with statistical indices such as root mean squared error (RMSE), the fraction of variance unexplained (FVU), and coefficient of determination (R2). Different simulations of performance assessment showed that the ANFIS techniques with requirements of past and current values of atmospheric pressures and height waves has more accuracy than the other techniques in the specified time and location. Meanwhile, in high lead times, the friction velocity decreases the accuracy of wave height forecasting.
