Towards a semantic-based automatic orchestration of geo-processing services

Many reasons may call for an orchestrated response computation to some geo service request: distributed data; specialized services (such as reprojection) are not available on the node addressed; heavy-weight jobs suggest distributed processing; and several more. Typically such orchestrations today are performed manually in advance. When it comes to the particularly interesting case of heavy-weight services, such as defined by the OGC Web processing service (WPS), it turns out very difficult at least to find methods for an automatic orchestration.We claim that this is due to the lack of explicit, machineunderstandable semantics of WPS. More promising seems to focus on an application profile of WPS, namely the OGC Web coverage processing service (WCPS), which offers a multidimensional raster processing language with a formal semantics; we believe that this language contains sufficient information for an automatic orchestration. In this contribution we address how to find ways to efficiently evaluate a WCPS request sent to a member of some resource-aware cloud where nodes have different service qualities, such as computational power, data available, or software installed.