pH acidification in the Red Sea: A machine learning-based validation study

This study presents application and performance comparison of various machine learning (ML) techniques to analyze pH variations in the Red Sea between the years 2021 and 2024, utilizing satellite remote sensing from the Copernicus Programme. The accuracy of the model is enhanced by employing data preprocessing. The performance of a number of machine learning models (Stepwise Linear Regression, Gaussian Process Regression, Linear Regression, Support Vector Machines and Neural Networks) are assessed. The results shown that the highest predictive accuracy is achieved by Stepwise Linear Regression and Linear Regression models. These models found to be superior in predicting pH changes due to seasonal phytoplankton blooms, vertical mixing of waters, and CO₂ infusion from the atmosphere accurately. Therefore, this research proposes a comprehensive approach for evaluating long-term changes in pH levels using robust data, improving strategic environmental governance in marine ecosystems. ML-based algorithms offer more integrated, cost-effective, and scalable solutions for monitoring ocean acidification, outperforming traditional approaches in both efficiency and adaptability. • Predicting ocean acidification using advanced ML models, providing critical insights for marine ecosystem management. • Models yielded more efficient and effective results compared to traditional methods • AI-based models offer a potential solution with a focus on environmental sustainability • AI-based applications offer significant potential for integration within remote monitoring systems