Spatial databases, optimized to store and query data representing objects defined in geometric space (e.g. address related data, GPS data, remote sensing data), can enable users to more effectively analyze and utilize location and geographic information. However, as geospatial datasets scale, it becomes increasingly challenging for these DBMS to provide interactive performance. Some of these databases use spatial approximations (mapping individual spatial objects to simpler geometries) to accelerate complex queries. However, spatial approximations are typically only used in an initial filtering step. Zacharatou et al. explore the application of approximate geospatial data processing techniques that do not require any exact geometric tests (but can still guarantee distance-based error bounds). They examine how GPUs that perform rasterization at interactive speeds can be leveraged to implement these techniques.