Earth sciences are about to follow the mapping domain where raster data increasingly get integrated in online services, contribute by far the largest volume. Interestingly, although more and more raster services are getting online, there is few work on a comprehensive model of raster services, rather architectures are of ad-hoc style and optimized towards very narrow applications, such as fast zoom and pan on seamless maps.
We claim that databases introduce a new quality of service on high-volume multi-dimensional earth science raster data, characterized by clear and understandable concepts, extensibility, and scalability. To support this, we present a comprehensive conceptual model for raster data in earth science and discuss how an efficient architecture can be derived from it, which is implemented in the rasdaman system. Further, we show how such concepts play a role in the development of OGC’s geo raster services, WCS and WCPS. Finally, we discuss some research challenges.