RAINFALL PREDICTION IN SURABAYA USING TABNET MODEL BASED ON RADIOSONDE DATA
DOI:
https://doi.org/10.36563/jkg10372Keywords:
rainfall prediction, TabNet, radiosonde data, tropical climate, deep learningAbstract
This study aims to predict daily rainfall in Surabaya using TabNet, a deep learning model for tabular data, based on radiosonde atmospheric parameters and official BMKG rainfall data. The radiosonde data include LI, KI, TT, CAPE, SWEAT, and SI measured at 00.00 and 12.00, while daily rainfall data were collected from a local station. All data were pre-processed to handle missing values, normalized, and used to construct relevant features for modeling. The dataset was divided into training and testing sets, and the TabNet model was trained to capture non-linear relationships between atmospheric parameters and rainfall. Model performance was evaluated using RMSE, MAE, MAPE, and R², achieving 6.58 mm, 3.14 mm, 17.69%, and 0.54, respectively, indicating good predictive ability. The predicted monthly rainfall for 2025 reflects seasonal patterns consistent with Surabaya’s tropical climate, with higher rainfall at the beginning and end of the year and lower values in mid-year months. The results demonstrate that TabNet can effectively capture seasonal rainfall trends, providing practical benefits for flood mitigation planning, water resource management, and early warning systems. This study highlights the potential of combining radiosonde data and deep learning for accurate rainfall prediction in tropical regions.
Downloads
References
Arik, S. O., & Pfister, T. (2021). TabNet: Attentive interpretable tabular learning. arXiv preprint arXiv:1908.07442. https://arxiv.org/abs/1908.07442
BMKG. (2022). Data curah hujan dan iklim Indonesia. Badan Meteorologi, Klimatologi, dan Geofisika. https://www.bmkg.go.id
Putra, A. D., & Lestari, S. (2022). Using radiosonde atmospheric parameters to predict convective rainfall in tropical regions. Journal of Atmospheric Research, 280, 106–119. https://doi.org/10.1016/j.atmosres.2022.106119
Nguyen, T., Tran, V., & Pham, H. (2023). Deep learning for climatological data: TabNet applications in rainfall prediction. Environmental Modelling & Software, 165, 105–127. https://doi.org/10.1016/j.envsoft.2023.105127
Yuan, L., Chen, Y., & Li, X. (2020). Deep learning approach for daily rainfall prediction using hydrometeorological datasets. Water, 12(7), 1945. https://doi.org/10.3390/w12071945
Hersbach, H., Bell, B., Berrisford, P., et al. (2020). The ERA5 global reanalysis. Quarterly Journal of the Royal Meteorological Society, 146(730), 1999–2049. https://doi.org/10.1002/qj.3803
Hermawan, I., Saputra, R., & Wulandari, D. (2021). Impact of vertical atmospheric data on rainfall prediction accuracy in tropical regions. International Journal of Climatology, 41(12), 5552–5567. https://doi.org/10.1002/joc.7185
Hermawan, I., Saputra, R., & Wulandari, D. (2021). Impact of vertical atmospheric data on rainfall prediction accuracy in tropical regions. International Journal of Climatology, 41(12), 5552–5567. https://doi.org/10.1002/joc.7185
Samet, R., & Shojaei, A. (2020). Mathematical modeling and optimization in civil engineering: A review. Journal of Civil Engineering and Management, 26(5), 417–432. https://doi.org/10.3846/jcem.2020.12635
Shang, Y., & Abourizk, S. M. (2018). Modeling construction project decision processes using digital simulations. Automation in Construction, 94, 75–84. https://doi.org/10.1016/j.autcon.2018.06.005
Sharma, A., & Tiwari, M. K. (2019). Linear programming techniques in construction management. Journal of Construction Engineering and Management, 145(2), 04018132. https://doi.org/10.1061/(ASCE)CO.1943-7862.0001586
Taha, H. A. (2017). Operations Research: An Introduction (10th ed.). Pearson Education.
Zavadskas, E. K., Turskis, Z., & Kildienė, S. (2016). Multi-criteria decision-making (MCDM) methods in civil engineering: Part I–A state-of-the-art survey. Archives of Civil and Mechanical Engineering, 16(1), 1–13. https://doi.org/10.1016/j.acme.2015.10.003
Downloads
Published
Issue
Section
License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
