CMIP5 기반 하천유량 예측을 위한 딥러닝 LSTM 모형의 최적 학습기간 산정
Estimation of Optimal Training Period for the Deep-Learning LSTM Model to Forecast CMIP5-based Streamflow
2
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
In this study, we suggested the optimal training period for predicting the streamflow using the LSTM (Long Short-Term Memory) model based on thedeep learning and CMIP5 (The fifth phase of the Couple Model Intercomparison Project) future climate scenarios. To validate the model performanceof LSTM, the Jinan-gun (Seongsan-ri) site was selected in this study. We comfirmed that the LSTM-based streamflow was highly comparable to themeasurements during the calibration (2000 to 2002/2014 to 2015) and validation (2003 to 2005/2016 to 2017) periods. Additionally, we compared theLSTM-based streamflow to the SWAT-based output during the calibration (2000∼2015) and validation (2016∼2019) periods. The results supportedthat the LSTM model also performed well in simulating streamflow during the long-term period, although small uncertainties exist. Then theSWAT-based daily streamflow was forecasted using the CMIP5 climate scenario forcing data in 2011∼2100. We tested and determined the optimaltraining period for the LSTM model by comparing the LSTM-/SWAT-based streamflow with various scenarios. Note that the SWAT-based streamflowvalues were assumed as the observation because of no measurements in future (2011∼2100). Our results showed that the LSTM-based streamflow wassimilar to the SWAT-based streamflow when the training data over the 30 years were used. These findings indicated that training periods more than30 years were required to obtain LSTM-based reliable streamflow forecasts using climate change scenarios.
Abstract
In this study, we suggested the optimal training period for predicting the streamflow using the LSTM (Long Short-Term Memory) model based on thedeep learning and CMIP5 (The fifth phase of the Couple Model Intercomparison Project) future climate scenarios. To validate the model performanceof LSTM, the Jinan-gun (Seongsan-ri) site was selected in this study. We comfirmed that the LSTM-based streamflow was highly comparable to themeasurements during the calibration (2000 to 2002/2014 to 2015) and validation (2003 to 2005/2016 to 2017) periods. Additionally, we compared theLSTM-based streamflow to the SWAT-based output during the calibration (2000∼2015) and validation (2016∼2019) periods. The results supportedthat the LSTM model also performed well in simulating streamflow during the long-term period, although small uncertainties exist. Then theSWAT-based daily streamflow was forecasted using the CMIP5 climate scenario forcing data in 2011∼2100. We tested and determined the optimaltraining period for the LSTM model by comparing the LSTM-/SWAT-based streamflow with various scenarios. Note that the SWAT-based streamflowvalues were assumed as the observation because of no measurements in future (2011∼2100). Our results showed that the LSTM-based streamflow wassimilar to the SWAT-based streamflow when the training data over the 30 years were used. These findings indicated that training periods more than30 years were required to obtain LSTM-based reliable streamflow forecasts using climate change scenarios.
참고문헌 (39)
- Abbaspour, K. C., "Swat-cup 2012. SWAT calibration and uncertainty program—A user manual", ~ (2012)
- Jeffrey G. Arnold, "A comprehensive surface-groundwater flow model", Journal of Hydrology, 142(1-4), pp.47~69 (1993), Elsevier BV[DOI]
- Georgy Ayzel, "Streamflow prediction in ungauged basins: benchmarking the efficiency of deep learning", E3S Web of Conferences, 163, pp.01001~ (2020), EDP Sciences
- T. Boulmaiz, "Impact of training data size on the LSTM performances for rainfall–runoffmodeling", Modeling Earth Systems and Environment, 6(4), pp.2153~2164 (2020), Springer Science and Business Media LLC[DOI]
- Bui, V., "Multi-behavior with bottleneck features LSTM for load forecasting in building energy management system", Electronics, 10(9), pp.1026~ (2021)
- Francesco Cioffi, "Large scale climate and rainfall seasonality in a Mediterranean Area: Insights from a non-homogeneous Markov model applied to the Agro-Pontino plain", Hydrological Processes, 31(3), pp.668~686 (2016), Wiley[DOI]
- Doğan, "Daily streamflow forecasting using artificial neural networks", International Congress on River Basin Management, pp.22~24 (2007)
- Minglei Fu, "Deep Learning Data-Intelligence Model Based on Adjusted Forecasting Window Scale: Application in Daily Streamflow Simulation", IEEE Access, 8, pp.32632~32651 (2020), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Shuai Gao, "Short-term runoffprediction with GRU and LSTM networks without requiring time step optimization during sample generation", Journal of Hydrology, 589, pp.125188~ (2020), Elsevier BV[DOI]
- Xinxin He, "Daily RunoffForecasting Using a Hybrid Model Based on Variational Mode Decomposition and Deep Neural Networks", Water Resources Management, 33(4), pp.1571~1590 (2019), Springer Science and Business Media LLC[DOI]
- Hinton, G., "Neural networks for machine learning lecture 6a overview of mini-batch gradient descent", Cited on, 14(8), pp.2~ (2012)
- Sepp Hochreiter, "Long Short-Term Memory", Neural Computation, 9(8), pp.1735~1780 (1997), MIT Press - Journals[DOI]
- Caihong Hu, "Deep Learning with a Long Short-Term Memory Networks Approach for Rainfall-RunoffSimulation", Water, 10(11), pp.1543~ (2018), MDPI AG[DOI]
- Dostdar Hussain, "A deep learning approach for hydrological time-series prediction: A case study of Gilgit river basin", Earth Science Informatics, 13(3), pp.915~927 (2020), Springer Science and Business Media LLC[DOI]
- Michael Hüsken, "Recurrent neural networks for time series classification", Neurocomputing, 50, pp.223~235 (2003), Elsevier BV[DOI]
- Patricia Jimeno-Sáez, "A Comparison of SWAT and ANN Models for Daily RunoffSimulation in Different Climatic Zones of Peninsular Spain", Water, 10(2), pp.192~ (2018), MDPI AG[DOI]
- 정성호, "딥러닝 기반 LSTM 모형을 이용한 감조하천 수위 예측", 한국수자원학회 논문집, 51(12), pp.1207~1216 (2018), 한국수자원학회[DOI]
- N. Kannan, "Sensitivity analysis and identification of the best evapotranspiration and runoffoptions for hydrological modelling in SWAT-2000", Journal of Hydrology, 332(3-4), pp.456~466 (2007), Elsevier BV[DOI]
- 김현일, "LSTM 모형과 로지스틱 회귀를 통한 도시 침수 범위의 예측", 대한토목학회논문집(국문), 40(3), pp.273~283 (2020), 대한토목학회[DOI]
- Frederik Kratzert, "Toward Improved Predictions in Ungauged Basins: Exploiting the Power of Machine Learning", Water Resources Research, 55(12), pp.11344~11354 (2019), American Geophysical Union (AGU)[DOI]
- Frederik Kratzert, "A note on leveraging synergy in multiple meteorological data sets with deep learning for rainfall–runoffmodeling", Hydrology and Earth System Sciences, 25(5), pp.2685~2703 (2021), Copernicus GmbH[DOI]
- Xuan-Hien Le, "River streamflow prediction using a deep neural network: a case study on the Red River, Vietnam", Korean Journal of Agricultural Science, 46(4), pp.843~856 (2019), 농업과학연구소[DOI]
- 이기하, "메콩강 유출모의를 위한 물리적 및 데이터 기반 모형의 비교 ․ 분석", 한국수자원학회 논문집, 51(6), pp.503~514 (2018), 한국수자원학회[DOI]
- MingXing Li, "Regional soil moisture simulation for Shaanxi Province using SWAT model validation and trend analysis", Science China Earth Sciences, 53(4), pp.575~590 (2010), Springer Science and Business Media LLC[DOI]
- Li, W., "High temporal resolution rainfall–runoffmodeling using long-short-termmemory(LSTM)networks", Neural Computing and Applications, 33, pp.1261~1278 (2021)[DOI]
- Darong Liu, "Streamflow Prediction Using Deep Learning Neural Network: Case Study of Yangtze River", IEEE Access, 8, pp.90069~90086 (2020), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Maas, A. L., "Rectifier nonlinearities improve neural network acoustic models", Proc. icml, 30(1), pp.3~ (2013)
- 마샤 모라디, "수문학적 예측을 위한 딥러닝기반 인공신경망의 최적화 알고리즘 비교: 남강댐 일유출량을 사례를 중심으로", 한국방재학회논문집, 18(6), pp.377~384 (2018), 한국방재학회[DOI]
- Pearson, K., "Correlation coefficient", Royal Society Proceedings, 58, pp.214~ (1895)
- Sandeep Samantaray, "Appraisal of RunoffThrough BPNN, RNN, and RBFN in Tentulikhunti Watershed: A Case Study", Frontiers in Intelligent Computing: Theory and Applications, pp.258~267 (2019), Springer Singapore[DOI]
- Shin, Y. C., "Analysis of watershed runoffand sediment characteristics due to spatio-temporal change in land uses using SWAT model", KCID Journal, 14(1), pp.50~56 (2007)
- Shin, Y. C., "Estimation of streamflow discharges using Kajiyama equation and SWAT Model", KCID Journal, 14(1), pp.41~49 (2007)
- 트란 광 카이, "딥러닝 기반 침수 수위 예측: 미국 텍사스 트리니티강 사례연구", 정보과학회논문지, 44(6), pp.607~612 (2017), 한국정보과학회
- Bhumika Uniyal, "Evaluation of SWAT simulated soil moisture at catchment scale by field measurements and Landsat derived indices", Agricultural Water Management, 193, pp.55~70 (2017), Elsevier BV[DOI]
- Zhongrun Xiang, "Distributed long-term hourly streamflow predictions using deep learning – A case study for State of Iowa", Environmental Modelling & Software, 131, pp.104761~ (2020), Elsevier BV[DOI]
- Zhongrun Xiang, "A Rainfall‐RunoffModel With LSTM‐Based Sequence‐to‐Sequence Learning", Water Resources Research, 56(1) (2020), American Geophysical Union (AGU)[DOI]
- Aijing Zhang, "Assessments of Impacts of Climate Change and Human Activities on Runoffwith SWAT for the Huifa River Basin, Northeast China", Water Resources Management, 26(8), pp.2199~2217 (2012), Springer Science and Business Media LLC[DOI]
- Jun Zhang, "Time series prediction using RNN in multi-dimension embedding phase space", Proc. 1998 IEEE International Conference on Systems, Man, and Cybernetics, 2, pp.1868~1873 (1998), IEEE
- Zhihua LV, "Predicting of RunoffUsing an Optimized SWAT-ANN: A Case Study", Journal of Hydrology: Regional Studies, 29, pp.100688~ (2020), Elsevier BV[DOI]