Buoy network is a promising technique to monitor air pollution in port water area, which is not limited by terrain and can provide continuous measurements at a low cost. Determining the number and location of buoys is a key issue in the design of the buoy network, however, it is affected by stochasticity of wind direction, wind speed, and emission intensity of ship exhaust. Therefore, a robust optimization approach is proposed to solve the problem of how to design buoy network in port water area for monitoring air pollution, i.e., to determine the number and location of buoys, with consideration of stochastic ship exhaust emissions and wind conditions. First, a bi-objective robust optimization model is developed based on the concept of p-robust to maximize monitoring capacity and minimize costs of buoy network. A method for constructing stochastic scenarios is subsequently proposed by combining Latin Hypercube Sampling, ship emission inventory, and atmospheric diffusion simulation. Then, a solution algorithm integrating Epsilon-constraint and Nash bargaining solution methods is presented to solve the model. Finally, a container port in eastern China is selected as a study case. Finally, the relative regret of the obtained robust layout is less than 13.04%. Results show that improving monitoring capacity and cross-scenario robustness is premised on increasing costs, decision-makers need to balance these aspects and design a buoy network with actual budgets. This study provides insights into how to design a buoy network to monitor air pollution in port water areas, and presents methods to determine the decision solution of a bi-objective robust optimization model, which helps maritime and environmental authorities manage ship emissions in port and coastal regions.