Journal of the Korea Academia-Industrial cooperation Society (한국산학기술학회논문지)

The Korea Academia-Industrial cooperation Society (한국산학기술학회)
권호 표지
DOI QR코드

DOI QR Code

Evaluation of LSTM Model for Inflow Prediction of Lake Sapgye

삽교호 유입량 예측을 위한 LSTM 모형의 적용성 평가

  • 황병기 (상명대학교 건설시스템공학과)
  • Received : 2021.01.06
  • Accepted : 2021.04.02
  • Published : 2021.04.30
Download PDF
⟨ Previous Next ⟩

Abstract

A Python-based LSTM model was constructed using a Tensorflow backend to estimate the amount of outflow during floods in the Gokgyo-cheon basin flowing into the Sapgyo Lake. To understand the effects of the length of input data used for learning, i.e., the sequence length, on the performance of the model, the model was implemented by increasing the sequence length to three, five, and seven hours. Consequently, when the sequence length was three hours, the prediction performance was excellent over the entire period. As a result of predicting three extreme rainfall events in the model verification, it was confirmed that an average NSE of 0.96 or higher was obtained for one hour in the leading time, and the accuracy decreased gradually for more than two hours in the leading time. In conclusion, the flood level at the Gangcheong station of Gokgyo-cheon can be predicted with high accuracy if the prediction is performed for one hour of leading time with a sequence length of three hours.

삽교호로 유입하는 곡교천 유역의 홍수시 유출량을 추정하기 위해서 Tensorflow를 활용하여 파이썬 기반의 LSTM 모형을 구축하였다. 층의 깊이가 성능에 미치는 영향을 분석하기 위해, 은닉층의 깊이를 2, 4, 6층으로 증가시키면서, 선행시간 1시간부터 5시간까지 예측을 수행하였으며, 은닉층의 개수가 4개일 때가 가장 우수한 성능을 나타내었다. 학습에 사용하는 입력자료의 길이 즉, 시퀀스길이가 모형의 성능에 미치는 영향을 파악하고자 시퀀스길이를 3시간, 5시간, 7시간으로 증가시키면서 모형을 실행한 결과, 시퀀스길이가 3시간 일 때, 전 시간대에 걸쳐 예측 성능이 우수한 것으로 분석되었다. 모형 검증에서 극한 강우 3건에 대하여 예측을 수행한 결과 선행시간 1시간에 대하여 평균 NSE 0.96 이상의 높은 정확도를 나타내었으며, 선행시간 2시간 이상에 대하여 정확도는 점차적으로 낮아지는 것으로 확인되었다. 결론적으로 시퀀스길이 3시간을 사용하여 선행시간 1시간에 대한 예측을 수행한다면 곡교천 강청 관측소의 홍수위를 높은 수준의 정확도로 예측할 수 있음을 확인하였다.

Keywords

References

  1. H. Jun, J. Lee, "A Methodology for Flood Forecasting and Warning Based on the Characteristic of Observed Water Levels Between Upstream and Downstream", Journal of Korea Society of Hazard Mitigation, Vol.13, No.6, pp.367-374, 2013. DOI: http://dx.doi.org/10.9798/KOSHAM.2013.13.6.367
  2. M. Moradi, T. Lee, "Comparison of Optimization Algorithms in Deep Learning-Based Neural Networks for Hydrological Forecasting: Case Study of Nam River Daily Runoff", Journal of Korea Society of Hazard Mitigation, Vol.18, No.6, pp.377-384, 2018. DOI: http://dx.doi.org/10.9798/KOSHAM.2018.18.6.377
  3. S. Jung, D. Lee, K. Lee, "Prediction of River Water Level Using Deep-Learning Open Library", Journal of Korea Society of Hazard Mitigation, Vol.18, No. 1, pp.1-11, 2018(a). DOI: http://dx.doi.org/10.9798/KOSHAM.2018.18.1.1
  4. S. Jung, H. Cho, J. Kim, G. Lee, "Prediction of Water Level in a Tidal River Using a Deep-learning Based LSTM Model", Journal of Korea Water Resources Association, Vol.51, No.12, pp.1207-1216, 2018(b). DOI: http://doi.org/10.3741/JKWRA.2018.51.12.1207
  5. M. K. Park, Y. S. Yoon, H. H. Lee, J. H. Kim, "Application of Recurrent Neural Network for Inflow Prediction into Multi-purpose Dam Basin", Journal of Korea Water Resources Association, Vol.51, No.12, pp.1217-1227, 2018. DOI: http://doi.org/10.3741/JKWRA.2018.51.12.1217
  6. D. Zhang, G. Lindholm, H. Ratnaweera, "Use Long Short-Term Memory to Enhance Internet of Things for Combined Sewer Overflow Monitoring", J. of Hydrology, Vol.556, pp.409-418, 2018(a). DOI: http://doi.org/10.1016/j.jhydrol.2017.11.018
  7. D. Zhang, E.K. Holland, G. Lindholm, H. Ratnaweera, "Hydraulic Modelling and Deep Learning Based Flow Forecasting for Optimizing Inter Catchment Wastewater Transfer", J. of Hydrology, Vol.557, pp. 792-802, 2018(b). DOI: http://doi.org/10.1016/j.jhydrol.2017.11.029
  8. Y. Qi, Z. Zhou, L. Yang, Y. Quan, Q. Mao, "A decomposition ensemble learning model based on LSTM neural network for daily reservoir inflow forecasting", Water Resources Management, 2019. DOI: https://doi.org/10.1007/s11269-019-02345-1
  9. X. He, J. Luo, G. Zuo, J. Xie, "Daily runoff forecasting using a hybrid model based on variational mode decomposition and deep neural networks", Water Resources Management, 33: 1571-1590, 2019. DOI: https://doi.org/10.1007/s11269-2183-x
  10. D. Zhang, N. Martinez, G. Lindholm, H. Ratnaweera, "Manage sewer in-line storage control using hydraulic model and recurrent neural network", Water Resources Management, 2018(c). DOI: https://doi.org/10.1007/s11269-018-1919-3
  11. I-F. Kao, Y. Zhou, L-C. Chang, F-J. Chang, "Exploring a long short-term memory based on encoder-decoder framework for multi-step ahead flood forecasting", J. of Hydrology, 2020. DOI: https://doi.org/10.1016/j.jhydrol.2020.124631
  12. S. Hochreuter and J. Schmidhuber, "Long Short-Term Memory", Neural Computation, Vol.9, No.8, :pp.1735-1780, 1997. DOI: https://doi.org/10.1162/neco.1997.9.8.1735
  13. Q.-K. Tran, S.-K. Song, "Water Level Forecasting based on Deep Learning: A Use Case of Trinity River-Texas-The United States", Journal of Korean Institute of Information Scientists and Engineers, Vol. 44, No.6, pp.607-612, 2017. DOI: http://doi.org/10.3741/JKWRA.2018.51.12.1207
  14. Ministry of Construction and Transportation(MOCT), "River Master Plan Report on Sapgyechun Watershed", 2012.
  15. Ministry of Environment(MOE), "Annual Report on Hydrological Survey", Geum River Flood Control Office, Gongju City, Chungnam Procince, 2019.
  16. Ministry of Land and Transportation(MOLT), "Water Resources Management Information System(WAMIS)", http://www.wamis.go.kr
  17. X.-H. Lee, H.V. Ho, G. Lee, S. Jung, "Application of Long Short-Term Memory (LSTM) Neral Network for flood forecasting", Water, 11, 2019. DOI: http://doi.org/10.3390/w11071387