Comparing a long short-term memory (LSTM) neural network with a physically-based hydrological model for streamflow forecasting over a Canadian catchment

被引:33
作者
Sabzipour, Behmard [1 ]
Arsenault, Richard [1 ]
Troin, Magali [1 ,2 ]
Martel, Jean-Luc [1 ]
Brissette, Francois [1 ]
Brunet, Frederic [1 ]
Mai, Juliane [3 ,4 ,5 ,6 ]
机构
[1] Univ Quebec, Ecole Technol Super, Hydrol Climate & Climate Change Lab, 1100 Notre Dame St West, Montreal, PQ H3C 1K3, Canada
[2] TVT Maison Numer & Innovat, Pl Georges Pompidou, F-83000 Toulon, France
[3] Univ Waterloo, Dept Civil & Environm Engn, 200 Univ Ave W, Waterloo, ON N2L 3G1, Canada
[4] Univ Waterloo, Dept Earth & Environm Sci, 200 Univ Ave W, Waterloo, ON N2L 3G1, Canada
[5] UFZ Helmholtz Ctr Environm Res, Dept Computat Hydrosyst, Permoser Str 15, D-04318 Leipzig, Germany
[6] Ctr Scalable Data Analyt & Artificial Intelligence, Humboldtstr 25, D-04105 Leipzig, Germany
基金
加拿大自然科学与工程研究理事会;
关键词
Long short-term memory (LSTM); Hydrological forecasting; Data assimilation; Ensemble forecasting; Deep learning; RAINFALL-RUNOFF MODELS; DATA ASSIMILATION; PREDICTION; DECOMPOSITION; BASINS; IMPACT;
D O I
10.1016/j.jhydrol.2023.130380
中图分类号
TU [建筑科学];
学科分类号
0813 ;
摘要
Streamflow forecasting is crucial in water planning and management. Physically-based hydrological models have been used for a long time in these fields, but improving forecast quality is still an active area of research. Recently, some artificial neural networks have been found to be effective in simulating and predicting short-term streamflow. In this study, we examine the reliability of Long Short-Term Memory (LSTM) deep learning model in predicting streamflow for lead times of up to ten days over a Canadian catchment. The performance of the LSTM model is compared to that of a process-based distributed hydrological model, with both models using the same weather ensemble forecasts. Furthermore, the LSTM's ability to integrate observed streamflow on the forecast issue date is compared to the data assimilation process required for the hydrological model to reduce initial state biases. Results indicate that the LSTM model forecasted streamflows are more reliable and accurate for lead-times up to 7 and 9 days, respectively. Additionally, it is shown that the LSTM model using recent observed flows as a predictor can forecast flows with smaller errors in the first forecasting days without requiring an explicit data assimilation step, with the LSTM model generating a median value of mean absolute error (MAE) for the first day of lead-time across all forecast issue dates of 25 m3/s compared to 115 m3/s for the assimilated hydrological model.
引用
收藏
页数:14
相关论文
共 59 条
  • [1] Forecasting daily river flows using nonlinear time series models
    Amiri, Esmail
    [J]. JOURNAL OF HYDROLOGY, 2015, 527 : 1054 - 1072
  • [2] Value of long-term streamflow forecasts to reservoir operations for water supply in snow-dominated river catchments
    Anghileri, D.
    Voisin, N.
    Castelletti, A.
    Pianosi, F.
    Nijssen, B.
    Lettenmaier, D. P.
    [J]. WATER RESOURCES RESEARCH, 2016, 52 (06) : 4209 - 4225
  • [3] Continuous streamflow prediction in ungauged basins: long short-term memory neural networks clearly outperform traditional hydrological models
    Arsenault, Richard
    Martel, Jean-Luc
    Brunet, Frederic
    Brissette, Francois
    Mai, Juliane
    [J]. HYDROLOGY AND EARTH SYSTEM SCIENCES, 2023, 27 (01) : 139 - 157
  • [4] Analysis of the effects of biases in ensemble streamflow prediction (ESP) forecasts on electricity production in hydropower reservoir management
    Arsenault, Richard
    Cote, Pascal
    [J]. HYDROLOGY AND EARTH SYSTEM SCIENCES, 2019, 23 (06) : 2735 - 2750
  • [5] On the Choice of Metric to Calibrate Time-Invariant Ensemble Kalman Filter Hyper-Parameters for Discharge Data Assimilation and Its Impact on Discharge Forecast Modelling
    Bergeron, Jean
    Leconte, Robert
    Trudel, Melanie
    Farhoodi, Sepehr
    [J]. HYDROLOGY, 2021, 8 (01) : 1 - 20
  • [6] Post-Processing of Stream Flows in Switzerland with an Emphasis on Low Flows and Floods
    Bogner, Konrad
    Liechti, Katharina
    Zappa, Massimiliano
    [J]. WATER, 2016, 8 (04)
  • [7] Hydro-economic assessment of hydrological forecasting systems
    Boucher, M. -A.
    Tremblay, D.
    Delorme, L.
    Perreault, L.
    Anctil, F.
    [J]. JOURNAL OF HYDROLOGY, 2012, 416 : 133 - 144
  • [8] Combining data assimilation and machine learning to emulate a dynamical model from sparse and noisy observations: A case study with the Lorenz 96 model
    Brajard, Julien
    Carrassi, Alberto
    Bocquet, Marc
    Bertino, Laurent
    [J]. JOURNAL OF COMPUTATIONAL SCIENCE, 2020, 44
  • [9] Impact of the quality of hydrological forecasts on the management and revenue of hydroelectric reservoirs - a conceptual approach
    Cassagnole, Manon
    Ramos, Maria-Helena
    Zalachori, Ioanna
    Thirel, Guillaume
    Garcon, Remy
    Gailhard, Joel
    Ouillon, Thomas
    [J]. HYDROLOGY AND EARTH SYSTEM SCIENCES, 2021, 25 (02) : 1033 - 1052
  • [10] Improving streamflow prediction in the WRF-Hydro model with LSTM networks
    Cho, Kyeungwoo
    Kim, Yeonjoo
    [J]. JOURNAL OF HYDROLOGY, 2022, 605