Prediction of Sea Surface Current Velocity and Direction Using LSTM

Irkhana Indaka Zulfa (Department of Mathematics, Sains and Technology UIN Sunan Ampel Surabaya)
Dian Candra Rini Novitasari (Department of Mathematics, Sains and Technology UIN Sunan Ampel Surabaya)
Fajar Setiawan (Perak Maritime Meteorology Station II, Surabaya)
Aris Fanani (Department of Mathematics, Sains and Technology UIN Sunan Ampel Surabaya)
Moh. Hafiyusholeh (Department of Mathematics, Sains and Technology UIN Sunan Ampel Surabaya)

Indonesian Journal of Electronics and Instrumentation Systems
Vol 11, No 1 (2021): April

 Labuan Bajo is considered to have an important role as a transportation route for traders and tourists. Therefore, it is necessary to have a further understanding of the condition of the waters in Labuan Bajo, one of them is sea currents. The purpose of this research is to predict sea surface flow velocity and direction using LSTM. There are many prediction methods, one of them is Long short-term memory (LSTM). The fundamental of LSTM is to process information from the previous memory by going through three gates, that is forget gate, input gate, and output gate so the output will be the input in the next process. Based on trials with several parameters namely Hidden Layer, Learning Rate, Batch Size, and Learning rate drop period, it achieved the smallest MAPE values of U and V components of 14.15% and 8.43% with 50 hidden layers, 32 Batch size and 150 Learn rate drop.