Assessing changes in the abundances of marine species in nearshore ecosystems on global scales is challenging due to the logistical difficulties of large-scale sampling. While aerial and satellite methods provide one solution, the resulting images often defy easy classification by computers, and require detailed evaluation by humans. While this is suitable for examining areas at sub-regional scales over a handful of years, it does not scale to global datasets over longer time periods. Citizen science is fast becoming a way to acquire large amounts of data from these kinds of imagery datasets that confound automated computer processing. We demonstrate the use of citizen science to create a 30-year dataset of global giant kelp cover. Here we present Floating Forests (http://floatingforests.org [2]): an international collaboration between kelp forest researchers and the citizen science organization Zooniverse. Floating Forests provides an interface that allows citizen scientists to identify canopy cover of giant kelp on Landsat images. This interface enables us to transform satellite imagery into kelp canopy data, and it serves as an outreach tool to engage and involve the public in coastal ecology. Floating Forests was launched on August 7, 2014; to date over 5,000 citizen scientists have classified over 500,000 Landsat images.