Military training areas generate complex mosaics of open landscapes including various disturbance dynamics that exhibit high biodiversity values. After the abandonment of military use, however, targeted habitat management measures have to be implemented to preserve open land that is valuable from a nature conservation perspective. Remote sensing makes an important contribution to the mapping and monitoring of habitat changes and management effects in ammunition-polluted areas. For this purpose, scale-specific machine learning methods are being developed for drone, aircraft and satellite-based sensor systems. Automated monitoring can be developed for multiple scales to record both, natural process dynamics and management success. In this way, early trends of spatio-temporal ecosystem dynamics can be made visible quantitatively and used to assess the nature conservation value of military training areas.