This presentation addresses the problem of coordinated Earth science campaign planning, the process of transforming a specification of the objectives of an Earth science campaign into a set of observations for accomplishing the campaign, utilizing diverse sensing resources from a collection of remote sensors. The paper also introduces a software architecture for a system that performs coordinated Earth science planning. The components of the architecture combine to allow for the formulation of campaign goals and plan activities, for automated or mixed initiative (human-in-loop) plan generation and execution, and dynamic replanning. The paper also provides illustrations of the campaign planning process based on a realistic Earth science scenario requiring multiple sensing resources. This example illustrates the challenges that need to be addressed in order to generate and execute campaign plans that optimally accomplish science objectives.