ESTIMATION OF LONGSHORE SEDIMENT TRANSPORT RATE, A COMPARISION BETWEEN SEMI EMPIRICAL FORMULAS AND ARTIFICIAL NEURAL NETWORK MODEL (ANN)

Measurement of the LSTR in the surf zone is one of the great challenges in coastal engineering and sciences. Several popular semi-empirical methods have beendeveloped to estimate the alongshore sediment transport rate (LSTR), including CERC Shore Protection Manual (1984), Kraus et al. (1989), Walton and Bruno (1989),Kamphuis (1991), Bayram et al. (2001), and Kumar et al. (2003) methods [1, 2, 3, 4, 5, 6]. Moreover, to estimate LST some methods based on artificial Neural NetworkModel (ANN) have been developed recently by some researchers. ANNs is a fairly new nonlinear statistical technique that can be used to solve problems that are not suitable for conventional statistical methods. Bakhtyar et al., by Neural Fuzzy inference system estimate the sediment transport rate in Arge coast of India [7]. Also,Hashemi et al. stated that the ANN could predict the seasonal changes in Tremadoc Bay coast profiles. In another studies, In addition [8], Kabiri - samani et al., byusing of Artificial Neural Network (ANN) and Fuzzy Logic (FL) evaluated the sediment transport rate methods in coastal zone of Iran and concluded that ANN, FL andGradient descent method have better efficiencies from the others [9]. In this case study, the ANN model is developed for estimation of alongshore sediment transport rate in Noorcoastal zone (Caspian Sea southern coasts) using wave parameters (wave breaking height (H), surf zone width (W) and alongshore current velocity (V)) as input and measuredsediment transport rate as output. The prediction accuracy of the trained ANN model is compared with the top three popular existing semi-empirical formulas, including CERC(1984), Walton and Bruno (1989), and Kamphuis (1991) methods for estimation of the LSTR [1,3,4].