TY - JOUR
T1 - Simulation of karst spring discharge using a combination of time–frequency analysis methods and long short-term memory neural networks
AU - An, Lixing
AU - Hao, Yonghong
AU - Yeh, Tian Chyi Jim
AU - Liu, Yan
AU - Liu, Wenqiang
AU - Zhang, Baoju
N1 - Publisher Copyright:
© 2020
PY - 2020/10
Y1 - 2020/10
N2 - Spring discharges from karst aquifers are results of spatially and temporally complex hydrologic processes, such as precipitation, surface runoff, infiltration, groundwater flow as well as anthropogenic factors. These processes are spatially and temporally varying at a multiplicity of scales with nonlinear and nonstationary characteristics. For improving the prediction accuracy of karst springs discharge, this study first applied the time–frequency analysis methods, including singular spectrum analysis (SSA) and ensemble empirical mode decomposition (EEMD) to extract frequency and trend feature of Niangziguan Springs discharge. Then the long short-term memory (LSTM) was used to simulate each frequency and trend subsequence. Subsequently, the prediction of spring discharge was completed by a combination of the simulated results from LSTM. Finally, the performances of LSTM, SSA-LSTM, and EEMD-LSTM under different inputs were compared. The results show that the performance of SSA-LSTM and EEMD-LSTM are better than LSTM, and the EEMD-LSTM model achieved the best prediction performance.
AB - Spring discharges from karst aquifers are results of spatially and temporally complex hydrologic processes, such as precipitation, surface runoff, infiltration, groundwater flow as well as anthropogenic factors. These processes are spatially and temporally varying at a multiplicity of scales with nonlinear and nonstationary characteristics. For improving the prediction accuracy of karst springs discharge, this study first applied the time–frequency analysis methods, including singular spectrum analysis (SSA) and ensemble empirical mode decomposition (EEMD) to extract frequency and trend feature of Niangziguan Springs discharge. Then the long short-term memory (LSTM) was used to simulate each frequency and trend subsequence. Subsequently, the prediction of spring discharge was completed by a combination of the simulated results from LSTM. Finally, the performances of LSTM, SSA-LSTM, and EEMD-LSTM under different inputs were compared. The results show that the performance of SSA-LSTM and EEMD-LSTM are better than LSTM, and the EEMD-LSTM model achieved the best prediction performance.
KW - Deep learning
KW - Ensemble empirical mode decomposition (EEMD)
KW - Karst spring discharge
KW - Long short-term memory (LSTM)
KW - Nonlinear and nonstationary time series
KW - Singular spectrum analysis (SSA)
UR - http://www.scopus.com/inward/record.url?scp=85089076277&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85089076277&partnerID=8YFLogxK
U2 - 10.1016/j.jhydrol.2020.125320
DO - 10.1016/j.jhydrol.2020.125320
M3 - Article
AN - SCOPUS:85089076277
SN - 0022-1694
VL - 589
JO - Journal of Hydrology
JF - Journal of Hydrology
M1 - 125320
ER -