Urban geospatial monitoring is a dynamic early warning in the process of urban development. It reflects urban spatial changes from a geospatial perspective. Modern urban digital governance must use spatial data infrastructures, integrate multisource data, and implement dynamic real-time monitoring in 3D space. To meet the real-time monitoring requirements of urban geographic space in 3D environments, the rapid construction of 3D scenes, the rapid processing of monitoring information, and dynamic process simulation have become key challenges. This paper presents an approach to urban geospatial monitoring that combines a sensor web and a high-performance computing infrastructure. The approach leverages Apache Spark and stream computing to transform the traditional processing algorithm into a data retrieval and analysis process supported by high-performance computing, which drives the rapid construction of 3D scenes and the dynamic calculation of sensor web observations. The effectiveness of this method is demonstrated through a case of urban waterlogging monitoring.