Abstract: Increasing shipping activity along Arctic routes has raised the risk of ship-source oil spills, while dynamic sea-ice conditions, complex oil-spill transport, and narrow emergency-response windows pose substantial challenges to conventional resource-scheduling approaches. This study proposes an emergency material scheduling framework that integrates dynamic environmental modeling with an improved multi-objective particle swarm optimization (IMOPSO) algorithm. A multi-objective optimization model is first formulated to simultaneously minimize emergency response time, total scheduling cost, and ecological risk while accounting for the dynamic evolution of sea ice and oil-spill dispersion. The conventional MOPSO algorithm is then enhanced through three mechanisms: an environmentally adaptive parameter-adjustment strategy to balance global exploration and local exploitation, an external archive maintenance strategy to improve the diversity and distribution of Pareto-optimal solutions, and a hybrid mutation operator combining Gaussian and Cauchy mutations to reduce premature convergence. A representative Arctic oil-spill scenario is simulated to evaluate the proposed method against the non-dominated sorting genetic algorithm Ⅱ (NSGA-Ⅱ). The results show that the proposed IMOPSO improves the hypervolume metric by 18.3% and the inverted generational distance metric by 33.3% relative to NSGA-Ⅱ, while achieving a more balanced trade-off among response time, economic cost, and ecological risk. Ablation experiments confirm the contribution of the individual algorithmic enhancements, and sensitivity analysis demonstrates robust performance under variations in environmental parameters. The proposed framework provides a quantitative decision-support approach for emergency resource allocation and response planning for ship-source oil spills in dynamic Arctic ice-covered waters.