Modern surveillance systems consist of multiple, geographically dispersed cameras, increasing the technical and scalability challenges for person re-identification. In this context, the use of geographical information to boost the effectiveness of a state-of-the-art re-identification algorithm has been implemented and evaluated, by leveraging the prediction of an event evolution. It is argued that the estimation of possible target trajectories can limit the footage search space and allow focused application of the re-identification algorithm. This is reflected in performance, effectiveness and scalability. The parametrization of the interesting footage reduction mechanism allows using different profiles and a flexible trade-off between performance and robustness. Our work is verified and evaluated in a well known benchmark dataset for re-identification and a real-world dataset created in the framework of the EU-project ADVISE.