A study on prediction of seawater intrusion in coastal aquifers in Korea using the LSTM model

Mincheol Kim; Jaehyun Bae; Woochang Jeong

doi:10.3741/JKWRA.2026.59.9.1049

All Issue

2026 Vol.59, Issue 9 Preview Page Next Page

A study on prediction of seawater intrusion in coastal aquifers in Korea using the LSTM model LSTM 모델을 이용한 국내 해안 지역 대수층 내 해수 침투 예측에 관한 연구

30 September 2026. pp. 1049-1059

PDF XML

Abstract

Seawater intrusion is a phenomenon that causes salinization of freshwater resources in coastal aquifers, and it is a key management concern in terms of the long-term stability of groundwater use and water resource management. However, since seawater intrusion is influenced by a combination of factors—including tides, precipitation, fluctuations in groundwater levels, and anthropogenic pumping—there are limitations to long-term predictions based solely on existing physics-based numerical models. In particular, the variability and uncertainty of field observation data are major factors that compromise the accuracy of seawater intrusion predictions. To address these limitations, this study aimed to predict the extent of seawater intrusion by utilizing long-term time-series data on electrical conductivity (EC) collected from groundwater monitoring stations in coastal areas of South Korea. By applying the LSTM (Long Short-Term Memory) model, which excels at learning long-term time-series dependencies, to the Haeun 1 and Mokdo 1 stations in the Gyeongnam region, the Gogum 1 and Sora 2 stations in the Jeonnam region, Daepo 1 and Joyang 1 in Gangwon Province; and Sungnoe 2 and Geunhong 2 in Incheon and South Chungcheong Province. After training the model on 70.0% of the total 3,437 time-series data points collected from January 1, 2014, to May 31, 2023, the model’s performance in predicting seawater intrusion was analyzed using the remaining 30.0%. Predictions made using an LSTM model based on long-term time-series EC data generally effectively captured the trends in seawater intrusion variability; however, differences in prediction performance were observed depending on the characteristics of variability by region and dataset. While the model demonstrated stable predictive performance in regions with relatively low volatility, it was found that prediction errors tended to increase at observation stations where sharp rises and falls occurred frequently.

Keywords

Seawater intrusion

Electrical conductivity

Time series forecasting

LSTM

해수 침투는 연안 대수층에서 담수 자원의 염분화를 유발하는 현상으로, 장기적인 지하수 이용 안정성과 수자원 관리 측면에서 중요한 관리 대상이다. 그러나 해수 침투 현상은 조석, 강수, 지하 수위 변동과 인위적 양수 등의 복합적인 요인에 의해 기존의 물리적 기반 수치모델만으로 장기 예측에 한계가 있다. 특히 현장 관측 자료의 변동성과 불확실성은 해수 침투 예측의 정확도를 저해하는 주요 요인으로 작용한다. 본 연구에서는 이러한 한계를 보완하기 위해 국내 해안 지역 지하수 관측소에서 장기간 측정된 전기전도도(Electrical Conductivity, EC) 시계열 자료를 활용하여 해수 침투 정도를 예측하고자 하였다. 장기 시계열 의존성 학습에 강점을 가지는 LSTM (Long Short-Term Memory) 모델을 적용하여 경남지역 해운 1과 목도 1, 전남지역 고금 1, 소라 2, 강원지역 대포 1, 조양 1 그리고 인천 및 충남지역 숭뇌 2, 근홍 2 관측소의 2014년 1월 1일부터 2023년 5월 31일까지 총 3,437개의 시계열 데이터 중 70.0 %를 학습시킨 후 30.0 %를 해수 침투 예측 특성을 분석하였다. 장기 시계열 EC 자료를 활용한 LSTM 모델을 통한 예측은 전반적으로 해수 침투 변동 경향을 효과적으로 모사하였으나, 지역별·자료별 변동성 특성에 따라 예측 성능 차이가 확인되었다. 변동성이 상대적으로 작은 지역에서는 안정적인 예측 성능을 보인 반면, 급격한 상승·하강이 빈번한 관측소에서는 예측 오차가 증가하는 경향이 확인되었다.

키워드

해수 침투

전기전도도

시계열예측

LSTM

References

Barlow, P.M., and Reichard, E.G. (2010). “Saltwater intrusion in coastal regions of North America.” Hydrogeology Journal, Vol. 18, pp. 247-260.

10.1007/s10040-009-0514-3

Bear, J., Cheng, A.H.-D., Sorek, S., Ouazar, D., and Herrera, I. (1999). Seawater intrusion in coastal aquifers: Concepts, methods and practices. Springer, Dordrecht, Netherlands.

10.1007/978-94-017-2969-7

Chen, C., Twycross, J., and Garibaldi, J.M. (2017). “A new accuracy measure based on bounded relative error for time series forecasting.” Plos One, Vol. 12, No. 3, pp. 1-23.

10.1371/journal.pone.017420228339480PMC5365136

Gers, F.A., Schmidhuber, J., and Cummins, F. (2000). “Learning to forget: Continual prediction with LSTM.” Neural Computation, Vol. 12, No. 10, pp. 2451-2471.

10.1162/089976600300015015

Ghyben, B.W. (1888). “Nota in verband met de voorgenomen putboring nabij Amsterdam (Notes on the probable results of well drilling near Amsterdam).” Tijdschrift van het Koninklijk Instituut van Ingenieurs, The Hague, Vol. 9, pp. 8-22.

Herzberg, A. (1901). “Die Wasserversorgung einiger Nordseebäder.” Journal of Gasbeleuchtung und Wasserversorgung, Vol. 44, pp. 815-819.

Jo, S.B., Jeon, B.C., Park, E.G., Choi, K.J., Song, S.H., and Kim, G.P. (2014). “Estimation of hydraulic characteristics and prediction of groundwater level in the eastern coastal aquifer of Jeju Island.” Journal of Environmental Science International, Vol. 23, No. 4, pp. 661-672.

10.5322/JESI.2014.4.661

Karim, F., Armin, M.A., Ahmedt-Aristizabal, D., Tychsen-Smith, L., and Petersson, L. (2023). “A review of hydrodynamic and machine learning approaches for flood inundation modeling.” Water, MDPI, Vol. 15, No. 3, 566.

10.3390/w15030566

Kassem, A., Sefelnasr, A., Kajla, N.I., Baig, F., and Sherif, M. (2026). “Dynamic prediction of seawater intrusion in hyper-arid coastal aquifer using advanced deep learning.” Earth Systems and Environment, Vol. 10, pp. 3769-3792.

10.1007/s41748-025-00792-y

Kim, K., Shin, J., Koh, E., Koh, G., and Lee, K. (2009). “Sea level rise around Jeju Island due to global warming and movement of groundwater/seawater interface in the eastern part of Jeju Island.” Journal of Soil and Groundwater Environment, Vol. 14, No. 3, pp. 68-79.

Korea Rural Community Corporation (KRC) (2002). Rural groundwater observation network system: Seawater intrusion warning, accessed 30 December 2022, <https://www.groundwater.or.kr/mw/seaWater/list.do>.

Lee, K.S., Kwak, S.H., and Lyu, S.W. (2012). “Analysis of saline wedge using 3D numerical model.” Proceedings of the Korea Water Resources Association Conference, pp. 571-575.

Lee, S., Kim, S., Jung, J., and Yoon, K.S. (2023). “Case study on flood water level prediction accuracy of LSTM model according to condition of reference hydrological station combination.” Journal of Korea Water Resources Association, Vol. 56, No. 12, pp. 981-992.

10.3741/JKWRA.2023.56.12.981

Li, M., Najjar, R.G., Kaushal, S., Mejia, A., Chant, R.J., Ralston, D.K., Burchard, H., Hadjimichael, A., Lassiter, A., and Wang, X. (2025). “The emerging global threat of salt contamination of water supplies in Tidal Rivers.” Environmental Science & Technology Letters, ACSPublications, Vol, 12, No. 8, pp. 881-892.

10.1021/acs.estlett.5c0050540822858PMC12351528

Maity, R., Srivastava, A., Sarkar, S., and Khan, M.I. (2024). “Revolutionizing the future of hydrological science: Impact of machine learning and deep learning amidst emerging explainable AI and transfer learning.” Applied Computing and Geosciences, Vol. 24, 100206.

10.1016/j.acags.2024.100206

Mosaffa, H., Sadeghi, M., Mallakpour, I., Jahromi, M.N., and Pourghasemi, H.R. (2022). “Application of machine learning algorithms in hydrology.” Computers in Earth and Environmental Sciences, Elsevier, Amsterdam, Netherlands, pp. 585-591.

10.1016/B978-0-323-89861-4.00027-0

Park, H.W. (2019). Research on LSTM performance in time series model prediction. Master Thesis, Keimyung University, pp. 13-23.

Park, J., Shin, J.-Y., Kim, S., and Kwon, J. (2025). “Assessing the applicability of the LTSF algorithm for streamflow time series prediction: Case studies of dam basins in South Korea.” Water, Vol. 17, No. 22, 3214.

10.3390/w17223214

Prinos, S.T., Wacker, M.A., Cunningham, K.J., and Fitterman, D.V. (2014). Origins and delineation of saltwater intrusion in the Biscayne aquifer and changes in the distribution of saltwater in Miami-Dade County. Florida. Scientific Investigations Report 2014-5025, U.S. Geological Survey, Reston, VA, U.S., p. 101.

10.3133/sir20145025

Song. G.D. (2024). A study on the prediction of water temperature using multivariate time-series data based on neural networks. Master Thesis, Busan University, pp. 29-44.

United Nations Office for Disaster Risk Reduction (UNDRR) (2026).Hazard Information profile: Saltwater intrusion (EN0302). accessed 16 May 2026, <https://www.undrr.org/understanding-disaster-risk/terminology/hips/en0302?utm_source=chatgpt.com>.

Werner, A.D. (2017). “On the classification of seawater intrusion.” Journal of Hydrology: Regional Studies, Vol. 551, pp 619-631.

10.1016/j.jhydrol.2016.12.012

Werner, A.D., Bakker, M., Post, V.E.A., Vandenbohede, A., Lu, C., Ataie-Ashtiani, B., Simmons, C.T., and Barry, D.A. (2013). “Seawater intrusion processes, investigation and management: Recent advances and future challenges.” Advances in Water Resources, Vol. 51, pp. 3-26.

10.1016/j.advwatres.2012.03.004

Wullems, B.J.M., Brauer, C.C., Baart, F., and Weerts, A.H. (2023). “Forecasting estuarine salt intrusion in the Rhine-Meuse delta using an LSTM model.” Hydrology and Earth System Sciences, EGU, No. 20, p. 3823-3850.

10.5194/hess-27-3823-2023

Yan, Y.-F., Liu, H.-X., Xu, S., Wang, Q.-L., Yang, Y.-H., Chen, Q.-Q., Wang, C.-Y., and Qin, T.-L. (2026). “Advances in coupling machine learning with hydrological simulation: A review.” Water Science and Engineering, Vol. 19, No. 1, pp. 1-10.

10.1016/j.wse.2026.01.002

Zhang, J., Zhu, Y., Zhang, X., Ye, M., and Yang, J. (2018). “Developing a long short-term memory (LSTM) based model for predicting water table depth in agricultural areas.” Journal of Hydrology, Vol. 561, pp. 918-929.

10.1016/j.jhydrol.2018.04.065

Zheng, R., Sun, Z., Jiao, J., Ma, Q., and Zhao, L. (2024). “Salinity prediction based on improved LSTM model in the qiantang estuary, China.” Journal of Marine Science and Engineering, Vol. 12, No. 8, 1339.

10.3390/jmse12081339

Information

Publisher :KOREA WATER RESOURECES ASSOCIATION
Publisher(Ko) :한국수자원학회
Journal Title :Journal of Korea Water Resources Association
Journal Title(Ko) :한국수자원학회 논문집
Volume : 59
No :9
Pages :1049-1059
Received Date : 2026-06-04
Revised Date : 2026-07-06
Accepted Date : 2026-07-06
DOI :https://doi.org/10.3741/JKWRA.2026.59.9.1049

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

All Issue