<p>Saltwater intrusion (SWI) poses a significant environmental challenge for coastal aquifers in Pacific Island nations, including Port Vila, Vanuatu. This study utilised a 3D numerical simulation model to evaluate SWI in the Tagabe coastal aquifer under current pumping regimes. To address SWI, optimal pumping patterns were identified through machine learning-based surrogate ensemble models and a simulation-optimisation (S–O) management model. A digital twin (DT) framework of the Tagabe coastal aquifer was developed, incorporating a 3D numerical model, surrogate ensemble models, and the S–O approach. The DT framework, linked with illustrative field data, was used to generate and analyse five illustrative scenarios based on varying salt concentrations (0.45, 0.55, 0.75, 0.90, and 1.15&#xa0;kg/m³; Scenarios 1 to 5, respectively). The results indicated that scenario 3 (salt concentration of 0.75&#xa0;kg/m³) led to the highest pumping rates from production wells (17,317 m³/d) and the lowest from barrier wells (202 m³/d), while scenario 5 showed maximum pumping of 31,676 m³/d from production wells and 5000 m³/d from barrier wells. The S–O model results were validated with less than 10% relative error compared to the numerical model outputs. To the author’s best knowledge, the application of DT in managing SWI has not been applied before. This study is the first to apply a DT framework for managing SWI in coastal aquifers, showcasing its potential for predicting future scenarios and optimising water management strategies. The results from the study indicated that DT can be successfully employed in a coastal aquifer for managing the SWI. The methodology developed and implemented in this study is of global significance and could be used to manage water resources wisely. The study demonstrated that with the help of the S–O approach, the DT is vital in predicting future scenarios, changes in pumping patterns, and other uncertainties.</p>