search for


Rainfall peak prediction using deep learning
Journal of the Korean Data & Information Science Society 2023;34:607-17
Published online July 31, 2023;
© 2023 Korean Data and Information Science Society.

Seunghawk Kim1 · Gwangseob Kim2 · Kyeong Eun Lee3

13Department of Statistics, Kyungpook National University
2Department of Civil Engineering
Correspondence to: This paper was conducted with the support of the Ministry of Public Administration and Security’s climate change response AI-based storm and flood risk prediction technology development project. (2022-MOIS61-002). This paper is a revised version of Seunghwak Kim’s Master’s Thesis.
1 Graduate student, Department of Statistics, Kyungpook National University, Daegu 41566, Korea.
2 Professor, Department of Civil Engineering, Kyungpook National University, Daegu 41566, Korea.
3 Associate professor, Department of Statistics, Kyungpook National University, Daegu 41566, Korea. E-mail:
Received June 26, 2023; Revised July 16, 2023; Accepted July 20, 2023.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited.
Due to the regional impact of climate change, the frequency and intensity of heavy rainfall events in South Korea have been increasing, while urbanization has contributed to a rise in internal urban flooding damages. Despite similar rainfall amounts, the characteristics of flood damages vary depending on the changes in the temporal distribution of rainfall. Therefore, a precise prediction of urban flooding requires a thorough understanding and prediction of the temporal distribution characteristics of rainfall. For this purpose, we utilized 48 years of data (from 1974 to 2021) from the end-point meteorological observation data provided by the meteorological data portal. We selected 60 regions and employed convolutional neural networks, long short-term memory networks, ConvLSTM, and EfficientnetV2 models to predict the rainfall peak locations within a 6-hour period for 11 urban areas prone to significant urban flooding. The analysis revealed that EfficientnetV2 achieved the highest accuracy (84.58%).
Keywords : Deep Learning, EfficientnetV2, Flooding, Rainfall Peak