Highlights: • Locusts are an old concern and swarms need to be controlled but they travel a lot. • A model of desert locust swarm migration was developed based on real observations. • With the possibility of flying upwind, our model better reproduces the migrations. • Food intake must be considered to regulate movement dynamics. Abstract: To be efficient, locust swarm control must focus on the place where eggs are laid and hopper bands may appear. But swarms travel a lot and among all the places likely to host them, there is a need to predict to which exactly they will fly. It is then essential to consider movement dynamics to anticipate any displacement that may lead to a further reproduction of locust swarms. Swarms mostly fly downwind and sometimes upwind. We designed an agent-based model to explore swarm displacements depending on the direction of the wind and the possibility for the swarms to realise upwind flights. A primary objective was to assess how upwind flights can improve the replicationand predictionof documented migratory paths. We looked at the effects of using upwind flight on the swarm ratio arriving in expected (i.e. historically known) areas. Our simulations clearly showed that using upwind flight helped for a better replication of Schistocerca gregaria migrations than not using upwind flight. Not using upwind flight reduced swarm dispersion and reduced the range of migrations. Hence, prevailing winds alone cannot explain locust swarm migrations. Food intake must also be considered to regulate movement dynamics and vegetated areas seem to be more attractive to locusts than expected. Our simulations did not perfectly reproduce the general patterns of migrations in some scenarios, but this invites further investigations and the use of other types of field data to calibrate the model. Nonetheless, our results highlighted the importance of upwind flight and showed the major role of wind and temperature on swarm displacement.