Dispersal is a ubiquitous phenomenon which affects the population dynamics and evolution of natural populations, hence it is a fundamental process in driving species persistence under global changes. However, dispersal rates and dispersal range are difficult to measure in most species, and remain unknown for many. In addition, informed behaviors, whereby individuals leave their natal area and select a breeding habitat non-randomly, may play an important role in species' responses to global change, and are even more difficult to comprehend. By developing new models combing demographic and genetic data, we reveal dispersal rates, range and behaviors for emperor penguin, a species threatened by climate change and living in the most remote place on earth where measurements of dispersal are nonexistent. We found that emperor penguins have a short distance of dispersal compared to their capacity to cover large distances during seasonal migration. On average, emigration rates are small, with emigration occurring when the local habitat becomes unsuitable (informed emigration). While overall dispersal rates remain low, occasional mass emigration events can occur in some regions. We also detect environmental and demographic drivers of emigration using new independent data:emigration is more likely to occur for habitats far from open water and with low food availability. Finally, we provided new global population forecasts for emperor penguins that incorporate dispersal processes into a sea-ice dependent demographic model that can inform conservation actions in Antarctica. Our method can be applied to other species and data-poor systems to estimate dispersal processes.