TY - JOUR
T1 - Hybrid harmonic analysis and wavelet network model for sea water level prediction
AU - El-Diasty, Mohammed
AU - Al-Harbi, Salim
AU - Pagiatakis, Spiros
N1 - Funding Information:
This project was funded by the Deanship of Scientific Research (DSR) , King Abdulaziz University, Jeddah, Saudi Arabia under grant no. RG-1-150-36 . The authors, therefore, acknowledge with thanks DSR technical and financial support. Also, the Monterey and Kiptopeke tide gauges data was obtained from NOAA, USA. The Ras-Tanoura and Jizan tide gauges data was provided by General Commission of Survey (GCS), Riyadh, KSA.
Publisher Copyright:
© 2017 Elsevier Ltd
PY - 2018/1
Y1 - 2018/1
N2 - Accurate sea water level prediction is required for safe marine navigation in shallow waters as well as for other marine operations. Traditionally, tide prediction is commonly carried out using only the harmonic analysis (HA-only) model or only a wavelet network (WN-only) model. The harmonic analysis method is the most reliable model for long term sea water level prediction when long data records are available and in contrast the wavelet network method is the most reliable model used for short term sea water level prediction when short data records are available. This paper developed a hybrid harmonic analysis and wavelet network (HA-and-WN) model for accurate sea water level prediction. To validate the hybrid HA-and-WN model, sea water level data from four tide gauges are employed to investigate the performance of the developed hybrid model. It is shown that the majority of error values at 95% confidence level fall within ±14.77 cm, ±2.65 cm and ±2.08 cm range in average with maximum error of 36.84 cm, 9.21 cm and 7.00 cm in average for HA-only model, WN-only model and hybrid HA-and-WN model, respectively. Also, it is found that the root-mean-squared (RMS) errors are about 9.75 cm, 1.85 cm and 1.49 cm for HA-only, WN-only and hybrid HA-and-WN models, respectively, based on the overall performance from the four tide gauges under implementation. Therefore, it is concluded that the developed hybrid HA-and-WN model is superior to the HA-only model by about 85% and outperforms the WN-only model by about 20%, based on the overall RMS errors.
AB - Accurate sea water level prediction is required for safe marine navigation in shallow waters as well as for other marine operations. Traditionally, tide prediction is commonly carried out using only the harmonic analysis (HA-only) model or only a wavelet network (WN-only) model. The harmonic analysis method is the most reliable model for long term sea water level prediction when long data records are available and in contrast the wavelet network method is the most reliable model used for short term sea water level prediction when short data records are available. This paper developed a hybrid harmonic analysis and wavelet network (HA-and-WN) model for accurate sea water level prediction. To validate the hybrid HA-and-WN model, sea water level data from four tide gauges are employed to investigate the performance of the developed hybrid model. It is shown that the majority of error values at 95% confidence level fall within ±14.77 cm, ±2.65 cm and ±2.08 cm range in average with maximum error of 36.84 cm, 9.21 cm and 7.00 cm in average for HA-only model, WN-only model and hybrid HA-and-WN model, respectively. Also, it is found that the root-mean-squared (RMS) errors are about 9.75 cm, 1.85 cm and 1.49 cm for HA-only, WN-only and hybrid HA-and-WN models, respectively, based on the overall performance from the four tide gauges under implementation. Therefore, it is concluded that the developed hybrid HA-and-WN model is superior to the HA-only model by about 85% and outperforms the WN-only model by about 20%, based on the overall RMS errors.
KW - Harmonic analysis method
KW - Neural networks
KW - Prediction
KW - Tide gauges
KW - Water levels
KW - Wavelet network
UR - http://www.scopus.com/inward/record.url?scp=85035063233&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85035063233&partnerID=8YFLogxK
U2 - 10.1016/j.apor.2017.11.007
DO - 10.1016/j.apor.2017.11.007
M3 - Article
AN - SCOPUS:85035063233
SN - 0141-1187
VL - 70
SP - 14
EP - 21
JO - Applied Ocean Research
JF - Applied Ocean Research
ER -