A probabilistic database can including a set of geospatially referenced records and a probability engine. The geospatially referenced records corresponding to a real-world volumetric space indexed against voxels. Each voxel can have a unique identifier in the probabilistic database. Each voxel can corresponds to a volumetric unit of the real-world volumetric space. Geospatial information referenced against voxels can be internally inconsistent, which results in a naturally probabilistic storage format. The probability engine can be operable to dynamically determine a certainty value for an object of the probabilistic database. The certainty value can be a value from zero to one hundred percent, which is based upon a statistical analysis of the internally inconsistent geospatial information referenced against the voxels. The internally inconsistent geospatial information can correspond to a volumetric unit of the real-world volumetric space in which the object has a probability of residing within.