We developed relationship between the marine carbonate system in the Changjiang Estuary and adjacent East China Sea shelf and both nutrient biogeochemistry and hydrography using canonical correspondence analysis (CCA), and estimated water column total alkalinity and pH through artificial neural networks (ANN). The CCA showed that carbonate parameters were strongly related to temperature (T), salinity (S), and dissolved oxygen (DO), which indicated both physical and biological processes had significant influence on the carbonate system. Accordingly, an artificial neural network was informed using matrices from measured parameters (pressure (Pre), T, S, DO, nitrate (N), phosphate (P), and silicate (Si)) during a shelf study in May 2017, which was applied to estimate water column pH and total alkalinity. Overall, the ANN model retrieved the variables with high accuracies (RMSE): 7.514 μmol/kg for alkalinity and 0.022 for pH. This was confirmed for the independent test set not include in the training process. The ANN model was also applied to produce high resolution total alkalinity and pH data using temperature, salinity and dissolved oxygen from July 2016 cruise. It is thus a promising method to derive distributions of key biogeochemical variables. Whilst our model is not presently informed to analyze ocean acidification, it will be used to inform on the seasonal and inter-annual variability of the carbonate system using historical ocean data where no carbonate measurements are available.