In order to address practical challenges like inadequate earthwork allocation refinement, underused multimodal transportation advantages, and uncertainty impacts, this paper proposes a grid management-based multimodal earthwork allocation optimization method for dredging projects. Unlike traditional dump truck-reliant earthwork projects, dredging projects involves complex water-land operations and can leverage land, water, and pipeline transportation collaboratively. Under earthwork excavation-filling balance constraints, the method considers machinery impacts on efficiency and feasibility, and incorporates excavation volume uncertainty. It aims to minimize total costs (transport machinery configuration, wharf construction, allocation, and delay penalties), constructing a two-stage stochastic programming model for grid-based multimodal allocation. The first stage decides machinery configuration and wharf location, while the second stage optimizes allocation plans for random scenarios. An efficient Benders decomposition algorithm is designed for solution, validated via a real dredging project. Results show the model cuts total transport costs significantly through grid-based fine management and multimodal collaboration, ensuring on-schedule completion. The algorithm outperforms Gurobi in solution quality and efficiency, providing reliable theoretical-methodological support for dredging project earthwork allocation planning and management.