An integrated deep learning approach for modeling dissolved oxygen concentration at coastal inlets based on...

Area of Science:

• Environmental Science
• Climate Change Research
• Coastal Oceanography

Background:

• Climate change significantly impacts dissolved oxygen (DO) concentrations in coastal areas.
• Predicting DO variation is complex due to numerous influencing water quality (WQ), hydrological, and climatic factors.
• Human activities in coastal inlets exacerbate DO challenges.

Purpose of the Study:

• To introduce and validate a Deep Learning Neural Network (DLNN) methodology for modeling and predicting DO concentrations.
• To assess the performance of conventional Machine Learning (ML) approaches against DLNN for DO prediction.
• To identify key hydroclimatic and WQ parameters influencing DO fluctuations in the Egyptian Rashid coastal inlet.

Main Methods:

• Utilized field-recorded WQ and hydroclimatic datasets from the Egyptian Rashid coastal inlet.
• Performed statistical and exploratory data analyses to understand DO fluctuation drivers.
• Employed conventional ML models (GPR, SVR, DTR) and a DLNN for DO prediction and comparison.

Main Results:

• DLNN achieved a 4% improvement in DO prediction accuracy over the best ML model.
• DLNN model demonstrated a high correlation (0.95) and low RMSE (0.42 mg/l).
• Solar radiation (SR), pH, water levels (WL), and atmospheric pressure (P) were identified as key influencing parameters.

Conclusions:

• DLNN offers superior accuracy for predicting DO concentrations in coastal environments.
• The developed models can serve as crucial indicators for coastal authorities monitoring climate change impacts.
• Effective DO monitoring is vital for managing coastal ecosystems under accelerated climate change.