All Issue

2022 Vol.64, Issue 1 Preview Page
Estimation of Optimal Training Period for the Deep-Learning LSTM Model to Forecast CMIP5-based Streamflow

CMIP5 기반 하천유량 예측을 위한 딥러닝 LSTM 모형의 최적 학습기간 산정

Beom-Seok Chuna
,
Tae-Hwa Leeb
,
Sang-Woo Kimc
,
Kyoung-Jae Limd
,
Young-Hun Junge
,
Jong-Won Dof
,
Yong-Chul Shing†
천 범석a
,
이 태화b
,
김 상우c
,
임 경재d
,
정 영훈e
,
도 종원f
,
신 용철g†
aMS Student, Department of Agricultural Civil Engineering, Kyungpook National University
bPh.D. Student, Department of Agricultural Civil Engineering, Kyungpook National University
cPh.D. Student, Department of Agricultural Civil Engineering, Kyungpook National University
dProfessor, Department of Rural Construction Engineering, Kangwon National University
eAssistant Professor, Department of Advanced Science and Technology Convergence, Kyungpook National University
fDeputy Director, Integrated Water Management Supporting Department, Rural Research Institute, Korea Rural Community Corporation
gAssociate Professor, Department of Agricultural Civil Engineering, Kyungpook National University
a경북대학교 농업토목공학과 MS학생
b박사 경북대학교 농업토목공학과 학생
c박사 경북대학교 농업토목공학과 학생
d강원대학교 농촌건설공학과 교수
e경북대학교 첨단과학기술융합학과 조교수
f차장, 통합물관리학과한국농어촌공사 농촌연구원
g경북대학교 농업토목공학과 부교수
† Corresponding author Tel.: +82-53-950-5734, Fax: +82-53-950-6752 E-mail: ycshin@knu.ac.kr
2022. pp. 39-50
PDF XML Citation
Abstract

In this study, we suggested the optimal training period for predicting the streamflow using the LSTM (Long Short-Term Memory) model based on the deep learning and CMIP5 (The fifth phase of the Couple Model Intercomparison Project) future climate scenarios. To validate the model performance of LSTM, the Jinan-gun (Seongsan-ri) site was selected in this study. We comfirmed that the LSTM-based streamflow was highly comparable to the measurements during the calibration (2000 to 2002/2014 to 2015) and validation (2003 to 2005/2016 to 2017) periods. Additionally, we compared the LSTM-based streamflow to the SWAT-based output during the calibration (2000~2015) and validation (2016~2019) periods. The results supported that the LSTM model also performed well in simulating streamflow during the long-term period, although small uncertainties exist. Then the SWAT-based daily streamflow was forecasted using the CMIP5 climate scenario forcing data in 2011~2100. We tested and determined the optimal training period for the LSTM model by comparing the LSTM-/SWAT-based streamflow with various scenarios. Note that the SWAT-based streamflow values were assumed as the observation because of no measurements in future (2011~2100). Our results showed that the LSTM-based streamflow was similar to the SWAT-based streamflow when the training data over the 30 years were used. These findings indicated that training periods more than 30 years were required to obtain LSTM-based reliable streamflow forecasts using climate change scenarios.

Keywords
Long Short-Term Memory (LSTM)
CMIP5
streamflow
deep learning
References

REFERENCES

1
K. C. Abbaspour, Swat-cup 2012. SWAT calibration and uncertainty program—A user manual (2013)
2
J. G. Arnold, P. M. Allen and G. Bernhardt, Journal of Hydrology, A comprehensive surface-groundwater flow model, 142(1-4); 47-69 (1993)10.1016/0022-1694(93)90004-s
3
G. Ayzel, L. Kurochkina, E. Kazakov and S. Zhuravlev, In E3S Web of Conferences, Streamflow prediction in ungauged basins: benchmarking the efficiency of deep learning, 163; 01001, EDP Sciences (2020)10.1051/e3sconf/202016301001
4
T. Boulmaiz, M. Guermoui and H. Boutaghane, Modeling Earth Systems and Environment, Impact of training data size on the LSTM performances for rainfall–runoff modeling, 6; 2153-2164 (2020)10.1007/s40808-020-00830-w
5
V. Bui, N. T. Le, V. H. Nguyen, J. Kim and Y. M. Jang, Electronics, Multi-behavior with bottleneck features LSTM for load forecasting in building energy management system, 10(9); 1026 (2021)
6
F. Cioffi, F. Conticello, U. Lall, L. Marotta and V. Telesca, Hydrological Processes, Large scale climate and rainfall seasonality in a Mediterranean Area: Insights from a non‐homogeneous Markov model applied to the Agro‐Pontino plain, 31(3); 668-686 (2017)10.1002/hyp.11061
7
Doğan , S. Işık, T. Toluk and M. Sandalcı, In International Congress on River Basin Management, Daily streamflow forecasting using artificial neural networks; 22-24 (2007)
8
M. Fu, T. Fan, Z. A. Ding, S. Q. Salih, N. Al-Ansari and Z. M. Yaseen, IEEE Access, Deep learning data-intelligence model based on adjusted forecasting window scale: Application in daily streamflow simulation, 8; 32632-32651 (2020)10.1109/access.2020.2974406
9
S. Gao, Y. Huang, S. Zhang, J. Han, G. Wang, M. Zhang and Q. Lin, Journal of Hydrology, Short-term runoff prediction with GRU and LSTM networks without requiring time step optimization during sample generation, 589; 125188 (2020)10.1016/j.jhydrol.2020.125188
10
X. He, J. Luo, G. Zuo and J. Xie, Water Resources Management, Daily runoff forecasting using a hybrid model based on variational mode decomposition and deep neural networks, 33(4); 1571-1590 (2019)10.1007/ s11269-019-2183-x.
11
G. Hinton, N. Srivastava and K. Swersky, Cited on, Neural networks for machine learning lecture 6a overview of mini-batch gradient descent, 14(8) (2012)
12
S. Hochreiter and J. Schmidhuber, Neural Computation, Long short-term memory, 9(8); 1735-1780 (1997)10.1162/ neco.1997.9.8.1735.
13
C. Hu, Q. Wu, H. Li, S. Jian, N. Li and Z. Lou, Water, Deep learning with a long short-term memory networks approach for rainfall-runoff simulation, 10(11); 1543 (2018)10.3390/w10111543
14
D. Hussain, T. Hussain, A. A. Khan, S. A. A. Naqvi and A. Jamil, Earth Science Informatics, A deep learning approach for hydrological time-series prediction: A case study of Gilgit river basin, 13(3); 915-927 (2020)10.1007/s12145-020-00477-2.
15
M. Hüsken and P. Stagge, Neurocomputing, Recurrent neural networks for time series classification, 50; 223-235 (2003)10.1016/S0925-2312(01)00706-8.
16
P. Jimeno-Sáez, J. Senent-Aparicio, J. Pérez-Sánchez and D. Pulido-Velazquez, Water, A comparison of SWAT and ANN models for daily runoff simulation in different climatic zones of peninsular Spain, 10(2); 192 (2018)10.3390/w10020192.
17
S. Jung, H. Cho, J. Kim and G. Lee, Journal of Korea Water Resources Association, Prediction of water level in a tidal river using a deep-learning based LSTM model, 51(12); 1207-1216, (in Korean) (2018)10.3741/ JKWRA.2018. 51.12.1207.
18
N. Kannan, S. M. White, F. Worrall and M. J. Whelan, Journal of Hydrology, Sensitivity analysis and identification of the best evapotranspiration and runoff options for hydrological modelling in SWAT-2000, 332(3-4); 456-466 (2007)10.1016/j.jhydrol.2006.08.001
19
H. I. Kim, K. Y. Han and J. Y. Lee, KSCE Journal of Civil and Environmental Engineering Research, Prediction of Urban Flood Extent by LSTM Model and Logistic Regression, 40(3); 273-283, (in Korean) (2020)10.12652/Ksce.2020.40.3.0273
20
F. Kratzert, D. Klotz, M. Herrnegger, A. K. Sampson, S. Hochreiter and G. S. Nearing, Water Resources Research, Toward improved predictions in ungauged basins: Exploiting the power of machine learning, 55(12); 11344-11354 (2019)10.1029/2019WR026065.
21
F. Kratzert, D. Klotz, S. Hochreiter and G. S. Nearing, Hydrology and Earth System Sciences, A note on leveraging synergy in multiple meteorological data sets with deep learning for rainfall–runoff modeling, 25(5); 2685-2703 (2021)10.5194/hess-25-2685-2021
22
X. H. Le, H. V. Ho and G. Lee, Korean Journal of Agricultural Science, River streamflow prediction using a deep neural network: a case study on the Red River, Vietnam, 46(4); 843-856 (2019)10.7744/kjoas.20190068
23
G. Lee, S. Jung and D. Lee, Journal of Korea Water Resources Association, Comparison of physics-based and data-driven models for streamflow simulation of the Mekong river, 51(6); 503-514, (in Korean) (2018)10.3741/JKWRA.2018.51.6.503
24
M. Li, Z. Ma and J. Du, Science China Earth Sciences, Regional soil moisture simulation for Shaanxi Province using SWAT model validation and trend analysis, 53(4); 575-590 (2010)10.1007/s11430-010-0031-1.
25
W. Li, A. Kiaghadi and C. Dawson, Neural Computing and Applications, High temporal resolution rainfall–runoff modeling using long-short-term- memory (LSTM) networks, 33; 1261-1278 (2021)10.1007/s11430-010-00 31-1.
26
D. Liu, W. Jiang, L. Mu and S. Wang, IEEE Access, Streamflow prediction using deep learning neural network: Case study of Yangtze river, 8; 90069-90086 (2020)10.1109/ACCESS.2020.2993874.
27
A. L. Maas, A. Y. Hannun and A. Y. Ng, In Proc, Rectifier nonlinearities improve neural network acoustic models, 30(1); 3 (2013)
28
M. Mahsa and T. Lee, Journal of the Korean Society of Hazard Mitigation, Comparison of optimization algorithms in deep learning-based neural networks for hydrological forecasting: Case study of nam river daily runoff, 18(6); 377-384, (in Korean) (2018)10.9798/kosham.2018.18.6.377
29
K. Pearson, In Royal Society Proceedings, Correlation coefficient, 58; 214 (1895)
30
S. Samantaray and A. Sahoo, In Frontiers in Intelligent Computing: Theory and Applications, Appraisal of runoff through BPNN, RNN, and RBFN in Tentulikhunti watershed: A case study; 258-267, Springer, Singapore (2020)10.1007/978-981-13-9920-6_26.
31
Y. C. Shin, K. J. Lim, K. S. Kim and J. D. Choi, KCID Journal, Analysis of watershed runoff and sediment characteristics due to spatio-temporal change in land uses using SWAT model, 14(1); 50-56, (in Korean) (2007)
32
Y. C. Shin, M. H. Shin, W. K. Kim, K. J. Lim and J. D. Choi, KCID Journal, Estimation of streamflow discharges using Kajiyama equation and SWAT Model, 14(1); 41-49, (in Korean) (2007)
33
Q. K. Tran and S. K. Song, Journal of KIISE, Water level forecasting based on deep learning: A use case of Trinity River- Texas-The United States, 44(6); 607-612, (in Korean) (2017)10.5626/JOK.2017.44.6.607.
34
B. Uniyal, J. Dietrich, C. Vasilakos and O. Tzoraki, Agricultural Water Management, Evaluation of SWAT simulated soil moisture at catchment scale by field measurements and Landsat derived indices, 193; 55-70 (2017)10.1016/j.agwat.2017.08.002
35
Z. Xiang and I. Demir, Environmental Modelling & Software, Distributed long-term hourly streamflow predictions using deep learning–A case study for State of Iowa, 131; 104761 (2020)10.1016/j.envsoft.2020.104761
36
Z. Xiang, J. Yan and I. Demir, Water Resources Research, A rainfall‐runoff model with LSTM‐based sequence‐to‐sequence learning, 56(1), e2019WR025326 (2020)10.1029/2019wr025326
37
A. Zhang, C. Zhang, G. Fu, B. Wang, Z. Bao and H. Zheng, Water Resources Management, Assessments of impacts of climate change and human activities on runoff with SWAT for the Huifa River Basin, Northeast China, 26(8); 2199-2217 (2012)10.1007/s11269-012-0010-8
38
Zhang and K. F. Man, In Proc. 1998 IEEE International Conference on Systems, Man, and Cybernetics, Time series prediction using RNN in multi-dimension embedding phase space., 2; 1868-1873, IEEE (1998)10.1109/ ICSMC.1998.728168.
39
L. V. Zhihua, J. Zuo and D. Rodriguez, Journal of Hydrology: Regional Studies, Predicting of runoff using an optimized SWAT-ANN: A case study, 29; 100688 (2020)10.1016/j.ejrh.2020.100688.
Information
  • Publisher :The Korean Society of Agricultural Engineers
  • Publisher(Ko) :한국농공학회
  • Journal Title :Journal of Korean Society of Agricultural Engineers
  • Journal Title(Ko) :한국농공학회논문집
  • Volume : 64
  • No :1
  • Pages :39-50
  • Received Date : 2021-08-24
  • Revised Date : 2021-12-13
  • Accepted Date : 2021-12-16
  • DOI :https://doi.org/10.5389/KSAE.2022.64.1.039
Journal Informaiton Journal of Korean Society of Agricultural Engineers Journal of Korean Society of Agricultural Engineers
  • NRF
  • KOFST
  • crossref crossmark
  • crossref cited-by
  • crossref funder-registry
  • crosscheck
  • orcid
  • open access
Journal Informaiton Journal Informaiton - close