EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Improving SST Prediction Accuracy of Ina-Flows Model in Karimun Jawa and Bira

Delfiana Y. Buti, Purwanti L. Sabrina, Adi Mulsandi, Avrionesti

JURNAL PESISIR DAN LAUT TROPIS · 2026

DOI: 10.35800/505qvc58 ↗

Kurz gefasst

Fachartikel in der Zeitschrift JURNAL PESISIR DAN LAUT TROPIS, 2026. Verfasst von Delfiana Y. Buti und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Sea surface temperature (SST) is an important parameter in ocean-atmosphere interaction, but BMKG Ina-Flows model outputs may still contain systematic bias. This study evaluates the accuracy of Ina-Flows SST predictions and improves their performance through machine learning-based bias correction at MAWS Karimun Jawa and Bira. The dataset consists of Ina-Flows SST outputs and MAWS observations from 1 December 2023 to 7 December 2024. Data up to 30 November 2024 were used for model training and validation, while 1–7 December 2024 was used as the test period. Three algorithms were compared: Support Vector Regression (SVR), Long Short-Term Memory (LSTM), and Bidirectional Long Short-Term Memory (Bi-LSTM). Model performance was evaluated using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Bias Error (MBE). The results show that Ina-Flows exhibits warm bias at both locations. At Karimun Jawa, LSTM produced the best performance with RMSE 0.207, MAE 0.182, and MBE -0.171, reducing RMSE by 29.83%. At Bira, Bi-LSTM performed best with RMSE 0.218, MAE 0.173, and MBE 0.011, reducing RMSE by 26.10%. These findings indicate that bias-correction effectiveness is location-dependent, so algorithm selection should be based on local validation. Keywords: bias correction, Ina-Flows, machine learning, SST, MAWS Abstrak Suhu permukaan laut (SST) merupakan parameter penting dalam interaksi laut-atmosfer, tetapi keluaran model Ina-Flows BMKG masih dapat mengandung bias sistematis. Penelitian ini bertujuan mengevaluasi akurasi prediksi SST Ina-Flows dan meningkatkan kinerjanya melalui koreksi bias berbasis machine learning pada lokasi MAWS Karimun Jawa dan Bira. Data yang digunakan berupa keluaran SST Ina-Flows dan observasi MAWS periode 1 Desember 2023 –7 Desember 2024. Data hingga 30 November 2024 digunakan untuk pelatihan dan validasi, sedangkan periode 1–7 Desember 2024 digunakan sebagai data uji. Tiga algoritma dibandingkan, yaitu Support Vector Regression (SVR), Long Short-Term Memory (LSTM), dan Bidirectional Long Short-Term Memory (Bi-LSTM). Kinerja model dievaluasi menggunakan Root Mean Square Error (RMSE), Mean Absolute Error (MAE), dan Mean Bias Error (MBE). Hasil menunjukkan bahwa Ina-Flows mengalami warm bias pada kedua lokasi. Di Karimun Jawa, LSTM memberikan kinerja terbaik dengan RMSE 0,207, MAE 0,182, dan MBE -0,171, serta menurunkan RMSE sebesar 29,83%. Di Bira, Bi-LSTM memberikan hasil terbaik dengan RMSE 0,218, MAE 0,173, dan MBE 0,011, serta menurunkan RMSE sebesar 26,10%. Temuan ini menunjukkan bahwa efektivitas algoritma koreksi bias bergantung pada karakteristik lokasi, sehingga pemilihan metode perlu disesuaikan melalui validasi lokal. Kata kunci: bias correction, Ina-Flows, machine learning, SST, MAWS

Zitieren

APA

Buti, D. Y., Sabrina, P. L., Mulsandi, A., & Avrionesti (2026). Improving SST Prediction Accuracy of Ina-Flows Model in Karimun Jawa and Bira. JURNAL PESISIR DAN LAUT TROPIS. https://doi.org/10.35800/505qvc58

Vancouver

Buti DY, Sabrina PL, Mulsandi A, Avrionesti. Improving SST Prediction Accuracy of Ina-Flows Model in Karimun Jawa and Bira. JURNAL PESISIR DAN LAUT TROPIS. 2026. doi:10.35800/505qvc58

BibTeX
@article{Buti2026,
  author = {Buti, Delfiana Y. and Sabrina, Purwanti L. and Mulsandi, Adi and Avrionesti},
  title = {Improving SST Prediction Accuracy of Ina-Flows Model in Karimun Jawa and Bira},
  journal = {JURNAL PESISIR DAN LAUT TROPIS},
  year = {2026},
  doi = {10.35800/505qvc58}
}
Kostenlos starten →