Optimal Experimental Design for Microplastics Sampling Experiments

Microplastics contamination is one of the most rapidly growing research topics. However, monitoring microplastics contamination in the environment presents both logistical and statistical challenges, particularly when constrained resources limit the scale of sampling and laboratory analysis. In this paper, we propose a Bayesian framework for the optimal experimental design of microplastic sampling campaigns. Our approach integrates prior knowledge and uncertainty quantification to guide decisions on how many spatial Centrosamples to collect and how many particles to analyze for polymer composition. By modeling particle counts as a Poisson distribution and polymer types as a Multinomial distribution, we developed a conjugate Bayesian model that enables efficient posterior inference. We introduce variance-based loss functions to evaluate expected information gain for both abundance and composition, and we formulate a constrained optimization problem that incorporates realistic cost structures. Our results provide principled and interpretable recommendations for allocating limited resources across the sampling and analysis phases. Through simulated scenarios and real-world-inspired examples, we demonstrate how the proposed methodology adapts to prior assumptions and cost variations, ensuring robustness and flexibility. This work contributes to the broader field of Bayesian experimental design by offering a concrete, application-driven case study that underscores the value of formal design strategies in environmental monitoring contexts.