Now, with the development of intelligent service technology, information provisioning models are changing from simple, single-type models to complex, on-demand models as required. Remote sensing analysis modeling often requires the use of multiple data sources, computer resources, tools, and models, and couples these resources into a work°ow for global change research. Therefore, remote sensing analysis modeling research today faces a critical challenge: the registry, discovery, and allocation of modeling resources are complex, as they are heterogeneous and geographically distributed. This paper proposes a resource package-oriented approach that provides guidelines for remote sensing analysis modeling to increase model sharing and integrating. This uni¯ed wrapping approach can be used to make the management and use of existing resources becoming easy and reusable. Based on this description of construction method, di®erent remote sensing analysis resources can be combined together on-demand. We developed a web environment to compose di®erent remote sensing modeling resources as ondemand processing components, based on the existing work°ow engineering, and use a dust storm monitoring model as an example. Comparative analyses are given to further explain the applications of this hierarchical construction method in the areas of the remote sensing analysis model.