We introduce a mobile spatial interactive application that uses a combination of a GPS, inertial sensing, gestural interaction, probabilistic models and Monte Carlo sampling, with vibration and audio feedback. This system allows the probing or querying of targets in a local area, based on a model of the local environment and specific context variables of interest, to enable a rich, embodied and location–aware spatial interaction. An experiment was conducted to investigate how spatial target selection at different distances, target separations and target widths is affected by a system with added ‘typical’ noise characteristics. Results showed that the successful selection of targets in the virtual environment is maximised with a combination of high angular separation and angular width.