For implementing dynamic aggregation of dispersive Geography Markup Language (GML) geospatial data, a aggregation mapping model based on GeoRSS standards was proposed considering the openness, self-description and dispersion characteristics of GML. A four-tier integrating framework of prototype architechture and its workflow were generated to utilize the above model to construct the related program. Then a aggregation prototype system was designed and developed for aggregating dispersive GML spatial data according to the system architechture. Based on this prototype system, the related experiments were performed to verify the feasibility of GML geospatial data aggregation, and confirm the correctness and availablity of the model. This prototype can help users to quikly find the GML geospatial data from the massive Geographic Information System (GIS) data, to achieve selection, analysis and classification of the existing data, as well as the instant updating and aggregating integration.