Modeling information resources by using the Topic Maps paradigm has the great benefit that each real world subject has to be represented as exactly one resource item of the data model. Therefore, Topic Maps stores have to guarantee that two information items representing the same subject, will be merged automatically. Because of that the model should never contains a subject twice. The main problem are the weak identities of current Topic Maps engines, which are often defined by humans and not by domain-specific algorithms or patterns. The special domain of spatial (geographical) knowledge provides such an algorithm based on a standardized geographical system, like the WGS-84. In this case, a Topic Maps engine can provide a powerful algorithm for unique identities of subjects and, moreover, geographical information can also be used for additional knowledge about the subject.