The abundance of freely available, scientific big data sets can facilitate discovery-based authentic science projects saving time, money, and effort. This is especially true of remotely sensed data, as there is global coverage of Earth's surface spanning several decades that can be used for a multitude of applications. In this article, I present three different case studies in which a project-based learning model was successfully integrated into undergraduate courses using big data to support authentic science. I illustrate this process, its implementation, and a timeline for use in both introductory and advanced undergraduate remote sensing courses. By participating in these projects, students learn the skills to link ground observation data with large, public-domain geospatial datasets to answer site- to landscape-level questions about the natural and built environment. In their multiscalar analysis of environmental data, students are forced to acknowledge the different yet overlapping operational scales of various social and ecological processes that drive landscape changes affecting Earth's resources. Student feedback from these courses has been positive, with participants indicating the projects gave them practical experience using geospatial technologies in real-world applications in natural resource management. From a teaching and learning perspective, the benefits of such an undertaking far outweigh the challenges, and I encourage others to consider a shift from traditional classroom practices to this more rewarding model of discovery.