It is important to accurately monitor coastal acidification (COA) in the coastal environment as this area is particularly affected by anthropogenic inputs. However, distributing pH sensors throughout Boston Harbor, Mass Bay or the Gulf of Maine with pH sensors would be impractical and very expensive. Computer simulations can help alleviate much of the cost and increase accurate predictions but requires validated, biochemical simulation models for the Northeast. We propose an integrated methodology that will combine and analyze all available information from monitoring stations in Mass Bay, Boston Harbor and their tributaries. We will use deep Gaussian Processe to blend information from diverse sources at variable fidelity in a multi-fidelity Bayesian framework, where all uncertainties are accounted for in the final prediction. Increased development of coastal communities has brought more runoff, sedimentation, nutrients and contaminants, and habitat degradation in coastal areas. Integrating multi-fidelity data with the biogeochemical-augmented coastal model (FVCOM-ERSEM) will result in highly accurate predictions of COA that will enable improved, cost-effective management of coastal resources, and support environmental preservation and remediation efforts.