یوسف کاظم زاده

Carbon sequestration is a key method for reducing atmospheric carbon dioxide (CO2), with ocean injection being particularly effective due to the oceans' capacity. However, this raises concerns about changes in water pH. Addressing the research gap in previous studies that utilized limited data, this study employs a large, diverse dataset for more accurate pH prediction under various environmental conditions. This study collected data to estimate the pH of pure water and brines saturated with CO2, including temperature, pressure, salinity, and pH. After ensuring proper data distribution and calculating statistical parameters, correlations between each input and the pH output were computed. Pressure had the highest correlation with pH, while temperature had the lowest and was the only parameter directly related to pH. The data was divided into training and testing sets, and a Gene Expression Programming model was developed and a simple correlation proposed. Various graphical and numerical methods evaluated the developed model. Error metrics for training and testing sets showed coefficient of determination values of 0.9037 and 0.9035, and root mean square error values of 0.1172 and 0.1182. Residual error plots indicated good performance across pressure, temperarure, salinity, and pH ranges. A cumulative frequency plot of absolute relative error showed all data points had errors below 0.102, with over 90 % below 0.0615. Kernel density estimation plots confirmed the model generally underestimated pH values. Sensitivity analysis using Spearman's correlation coefficients confirmed pressure had the highest and temperature had the lowest influence on pH output. The agreement between these coefficients and experimental data shows the model accurately mimics natural behaviors. This model provides valuable insights for predicting pH changes in CO2-saturated solutions, essential for carbon sequestration and related environmental studies. Understanding these changes helps assess