Development of a cluster analysis-based heavy rainfall damage prediction functions and their performance comparison

Jin Gyeong Yoo; Donghwi Jung; Do Guen Yoo

doi:10.3741/JKWRA.2025.58.11.993

All Issue

2025 Vol.58, Issue 11 Preview Page Next Page

Research Article

Development of a cluster analysis-based heavy rainfall damage prediction functions and their performance comparison 군집분석 기반 호우피해예측 함수 개발 및 예측 성능 비교

30 November 2025. pp. 993-1003

PDF XML

Abstract

According to the Ministry of the Interior and Safety's Disaster Yearbook, heavy rain continues to cause damage in South Korea every year, and property damage accounts for the largest proportion of natural disasters. Therefore, this study classified local governments in Gyeonggi Province using K-means clustering analysis, considering regional characteristics such as rainfall intensity, geographic location, and damage scale. Based on this, a heavy rain damage prediction function was developed. The analysis focused on local governments in Gyeonggi-do, using data from 2013 to 2023. Predictor variables consisted of meteorological and socioeconomic factors. Decision Tree Bagging (DT-Bagging) and Long Short-Term Memory (LSTM) models, which have demonstrated excellent predictive performance in previous studies, were used as prediction models. Mean Square Error (MSE), Root Mean Square Error (RMSE), and Normalized Root Mean Squared Error (NRMSE) were used to compare the predictive performance of the two models. The prediction performance comparison results showed that the DT-Bagging function achieved a 97.2% improvement over the LSTM function. Therefore, this study is expected to contribute to the establishment of a precise flood damage prediction and disaster management system based on the characteristics of local governments.

Keywords

Heavy rainfall

Damage prediction

Clustering

Predictive performance

행정안전부 재해연보에 따르면 우리나라에서는 매년 호우로 인한 피해가 지속되며, 자연재해 중 재산피해 비중이 가장 높다. 이에 본 연구는 강우 규모, 지리적 위치, 피해 규모 등과 같은 지역적 특성을 고려하여 군집분석(K-means)을 통해 경기도 내 지자체를 분류하고, 이를 바탕으로 호우피해예측 함수를 개발하였다. 분석 대상은 경기도 내 지자체이며, 2013년부터 2023년까지의 데이터를 활용하였다. 예측 변수는 기상요소와 사회·경제적 요인으로 구성하였다. 예측 모델로는 선행연구에서 예측 성능이 우수했던 의사결정나무 기반 배깅(Decision Tree Bagging, DT-Bagging)과 장단기메모리(Long Short-Term Memory, LSTM)을 적용하였다. 두 가지 모델의 예측 성능 비교를 위해 평균제곱오차(Mean Square Error, MSE), 평균제곱근오차(Root Mean Square Error, RMSE), 정규화 평균제곱근오차(Normalized Root Mean Squared Error, NRMSE)를 사용하였다. 예측 성능 비교 결과, DT-Bagging 함수가 LSTM 함수 대비 97.2%의 향상률을 보였다. 따라서 본 연구는 지자체 특성에 기반한 정밀한 호우피해예측과 재난관리 체계 구축에 기여할 수 있을 것으로 기대된다.

키워드

호우

피해 예측

군집분석

예측 성능

References

Amin, M.N., Iftikhar, B., Khan, K., Javed, M.F., AbuArab, A.M., and Rehman, M.F. (2023). “Prediction model for rice husk ash concrete using AI approach: Boosting and bagging algorithms.” Structures, Vol. 50, pp. 745-757.

10.1016/j.istruc.2023.02.080

Bhattarai, R., Yoshimura, K., Seto, S., Nakamura, S., and Oki, T. (2016). “Statistical model for economic damage from pluvial floods in Japan using rainfall data and socioeconomic parameters.” Natural Hazards and Earth System Sciences, Vol. 16, pp. 1063-1077.

10.5194/nhess-16-1063-2016

Choi, C.H., Park, K.H., Park, H.K., Lee, M.J., Kim, J.S., and Kim, H.S. (2017). “Development of heavy rain damage prediction function for public facility using machine learning.” Journal of the Korean Society of Hazard Mitigation, Vol. 17, No. 6, pp. 443-450.

10.9798/KOSHAM.2017.17.6.443

Hochreiter, S., and Schmidhuber, J. (1997). “Long short-term memory.” Neural Computation, Vol. 9, No. 8, pp. 1735-1780.

10.1162/neco.1997.9.8.1735

Kabari, L.G., and Onwuka, U.C. (2019). “Comparison of Bagging and Voting ensemble machine learning algorithm as a classifier.” International Journal of Advanced Research in Computer Science and Software Engineering, Vol. 9, No. 3, pp. 19-23.

Korea Research Institute for Human Settlements (KRIHS) (2022). Countermeasures for urban flood prevention in the era of climate crisis: Lessons from the 2022 metropolitan heavy rainfall in Korea, p. 3.

Lee, J.H. (2024). Machine learning model development for predicting heavy rain damage of grouped regions by regional characteristics. Master Thesis, Inha University.

Ministry of the Interior and Safety (MOIS) (2023). Disaster yearbook 2023. p. 12.

Moon, G.H. (2023). Decision of water quality related priority control district metered area for water distribution networks using cluster analysis. Master Thesis, The University of Suwon, p. 31.

Papagiannaki, K., Kotroni, V., Lagouvardos, K., Bezes, A., Vafeiadis, V., Messini, I., Kroustallis, E., and Totos, I. (2022). “Identification of rainfall thresholds likely to trigger flood damages across a Mediterranean region, based on insurance data and rainfall observations.” Water, Vol. 14, pp. 1-16.

10.3390/w14060994

Song, Y., Lee, H.J., Joo, J., and Park, M. (2024). “Predicting damage characteristics of heavy rain disasters using artificial neural network.” Journal of the Korean Society of Hazard Mitigation, Vol. 24, No. 1, pp. 83-89.

10.9798/KOSHAM.2024.24.1.83

Wang, H., Xu, S., Xu, H., Wu, Z., Wang, T., and Ma, C. (2023). “Rapid prediction of urban flood based on disaster-breeding environment clustering and Bayesian optimized deep learning model in the coastal city.” Sustainable Cities and Society, Vol. 99, pp. 1-16.

10.1016/j.scs.2023.104898

Wang, M., Fan, H., Yuan, H., Zhang, D., Su, J., Zhou, S., Zhang, Q., and Li, J. (2024). “Urban flooding damage prediction in matrix scenarios of extreme rainfall using a convolutional neural network.” Journal of Hydrology, Vol. 644, pp. 1-13.

10.1016/j.jhydrol.2024.132069

Information

Publisher :KOREA WATER RESOURECES ASSOCIATION
Publisher(Ko) :한국수자원학회
Journal Title :Journal of Korea Water Resources Association
Journal Title(Ko) :한국수자원학회 논문집
Volume : 58
No :11
Pages :993-1003
Received Date : 2025-08-25
Revised Date : 2025-10-02
Accepted Date : 2025-10-10
DOI :https://doi.org/10.3741/JKWRA.2025.58.11.993

Journal of Korea Water Resources Association ISSN:2799-8746(Print) 2799-8754(Online) 한국수자원학회 논문집

All Issue