We present a large-scale study on Geospatial Big Data Analytics in a festival management and crowd safety scenario based on our volunteer work at the largest music festival in Northern Europe, the 2017 and 2018 Roskilde music festival. As large crowds move between concerts, campsites, private parties and public viewing of the FIFA world cup soccer matches across the vast festival area, previously available visualization solutions for the crowd safety staff at Roskilde Festival lack a real-time visualization of crowd trajectory for monitoring of previously known chokepoints and the discovery of potential future chokepoints. We present a real-time visualization of crowd trajectory based on mobile device GPS data that is collected through the festival’s smartphone app from a significant subset of all festivalgoers. Hence, we present a recent case for the applications of real-time visualization of geospatial data for large-scale open-air events outside the confinements of urban architecture. Furthermore, we investigate and quantify the phenomenon of festivalgoers staying at the camping areas and sometimes not once visiting the musical performances in the inner perimeters of the festival, thereby finding that up to 3,000 festivalgoers both in 2017 and 2018 chose to not attend musical performances at all. Subsequently, we evaluate a Facebook-based approach on concert audience prediction based on social media audience overlaps between artist and festival Facebook pages, and benchmark its predictive power with a GPS-based measurement of audience sizes based on mobile device GPS data.