TEST OF LSTM NETWORKS IN LONG-TERM BEACH MORPHOLOGICAL CHANGES

In the prediction of beach profile changes, for example, long-term calculation daily changes can only provide sufficiently reliable reproduction results a few years. One reasons is that actual morphological change caused by superposition complex processes are unknown. The timeseries data data-driven models in contrast to physical has been applied various fields recent years with spread deep learning and expected be used as tentative solution problems cannot adequately predicted modeling. Recurrent neural networks being predict shoreline over several (Montaño et al., 2020), but application spatial such rarely investigated due limitations large amount observation required learning. Here, we LSTM (Long Short-Term Memory) network, one recurrent networks, observed at Hasaki coast, Japan changes.