Model Prediksi Kecepatan Arus Laut Di Perairan Selat Sunda Menggunakan Long Short-Term Memory (LSTM)

Daud, Anton (2025) Model Prediksi Kecepatan Arus Laut Di Perairan Selat Sunda Menggunakan Long Short-Term Memory (LSTM). Masters thesis, Institut Teknologi Sepuluh November.

Text 6016232038-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (1MB) | Request a copy

Abstract

Perairan Selat Sunda memiliki dinamika arus laut yang kompleks akibat interaksi berbagai faktor oseanografi seperti pasang surut, angin, dan perubahan iklim musiman. Pemahaman terhadap pola kecepatan arus laut sangat penting untuk mendukung aktivitas navigasi, perikanan, dan mitigasi bencana laut. Penelitian ini bertujuan memprediksi kecepatan arus laut di Selat Sunda menggunakan pendekatan Deep Learning dengan model Long Short-Term Memory (LSTM). Model dikembangkan berdasarkan data historis kecepatan arus laut periode 2022– 2024 yang diperoleh melalui observasi lapangan dan data sekunder. Proses penelitian meliputi pengolahan data, normalisasi, pelatihan model LSTM, dan evaluasi performa menggunakan metrik Mean Squared Error (MSE), Mean Absolute Error (MAE), dan R² Score. Hasil menunjukkan model memiliki performa baik dengan nilai evaluasi kuantitatif berupa MSE sebesar 186.71 (cm/s2), MAE sebesar 9.06 (cm/s), dan R² sebesar 0,868, yang menunjukkan model dapat menjelaskan sekitar 86,8% variasi data aktual. Model selanjutnya diimplementasikan ke dalam aplikasi web sederhana untuk menyajikan hasil prediksi secara real-time berdasarkan input pengguna. Evaluasi kegunaan aplikasi dilakukan menggunakan metode System Usability Scale (SUS) terhadap sepuluh responden. Skor rata-rata SUS sebesar 82,5 menunjukkan bahwa aplikasi berada dalam kategori “Excellent”, menandakan penerimaan yang sangat baik dari sisi antar muka, efisiensi, dan kemudahan penggunaan. Penelitian ini menyimpulkan bahwa pendekatan LSTM efektif untuk prediksi kecepatan arus laut dan telah berhasil diterapkan secara praktis dalam bentuk aplikasi web. Pengembangan lebih lanjut disarankan melalui integrasi variabel oseanografi tambahan, optimasi model, serta penyempurnaan antarmuka untuk mendukung pengambilan keputusan di bidang kelautan.
====================================================================================================================================
The Sunda Strait waters have complex ocean current dynamics due to the= interaction of various oceanographic factors such as tides, winds, and seasonal climate change. Understanding ocean current speed patterns is very important to support navigation, fisheries, and marine disaster mitigation activities. This study aims to predict ocean current speed in the Sunda Strait using the Deep Learning approach with the Long Short-Term Memory (LSTM) model. The model was developed based on historical ocean current speed data for the period 2022–2024 obtained through field observations and secondary data. The research process includes data processing, normalization, LSTM model training, and performance evaluation using the Mean Squared Error (MSE), Mean Absolute Error (MAE), and R² Score metrics. The results show that the model has good performance with quantitative evaluation values in the form of MSE of 186.71 (cm/s2), MAE of 9.06 (cm/s), and R² of 0.868, which indicates that the model can explain around 86.8% of the variation in actual data. The model is then implemented into a simple web application to present real-time prediction results based on user input. The usability evaluation of the application was conducted using the System Usability Scale (SUS) method on ten respondents. The average SUS score of 82.5 indicates that the application is in the “Excellent” category, indicating very good acceptance in terms of interface, efficiency, and ease of use. This study concludes that the LSTM approach is effective for predicting ocean current speed and has been successfully implemented practically in the form of a web application. Further development is suggested through the integration of additional oceanographic variables, model optimization, and interface refinement to support decision making in the marine sector.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Arus Laut; LSTM; Selat Sunda; Time Series. LSTM, Ocean Currents; Sunda Strait, Time Series.
Subjects:	G Geography. Anthropology. Recreation > G Geography (General) > G70.5.I4 Remote sensing G Geography. Anthropology. Recreation > GC Oceanography
Divisions:	Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Geomatics Engineering > 29101-(S2) Master Thesis
Depositing User:	Anton Daud
Date Deposited:	24 Jul 2025 06:04
Last Modified:	24 Jul 2025 06:04
URI:	http://repository.its.ac.id/id/eprint/119507

Actions (login required)

View Item