Hydrological models, as critical tools for water resource management, flood prediction, and ecosystem simulation, rely heavily on efficient sharing and collaboration to advance scientific research and engineering applications. Traditional hydrological models, often developed in standalone environments, face challenges such as data silos, cumbersome collaboration workflows, and coarse-grained permission management. While cloud computing has enabled the migration of hydrological models to cloud platforms, two core challenges persist in multi-user collaboration: (1) the complex hierarchical dependencies of model and data resources, which require data integrity during sharing, and (2) the need for fine-grained permission design to balance openness and security. This paper addresses these issues by proposing a cloud-based collaborative sharing method for hydrological models that integrates multi-user, multi-level, and multi-permission mechanisms. By establishing a dual-role system (individual users and administrators), categorizing data resources into five hierarchical levels (modeling data, input data, parameter schemes, scenario schemes, and simulation results), and defining three permission mechanisms (usage, co-construction, and backup rights), the method achieves efficient sharing and secure control of models. The framework supports dynamic model sharing, bookmarking, and backup while ensuring data dependency integrity. Empirical validation demonstrates significant improvements: collaborative task completion time is reduced. This method provides technical support for cloud-based transformation of hydrological models and offers new insights for cross-domain model sharing and collaborative innovation.