One of the most promising climate mitigation and adaptation interventions in agroecosystems is the practice of cover cropping. In temperate climates, such as the U.S. Midwest, cover cropping occurs during winter when growing conditions for vegetation is minimized. Winter cover crops are crucial to climate mitigation because of their potential to dramatically increase soil carbon sequestration. Our ability to monitor cover crop adoption and predict the benefits of widespread cover crop use as a potential climate mitigation strategy depends on our ability to accurately detect cover crops across regional scales. In this paper we evaluate the predictive capacity of established and novel remote-sensing-based metrics for detecting cover crops using a series of machine learning models. The goal of this work is to assess which regions of the electromagnetic spectrum and indices are most effective for detecting cover crops and crop residues.