Improving Efficiency of Grid Representation in GML

Abstract

Earth observation and remote sensing today are a primary resource for a miscellanea of geoservices providing massive timeseries of raster-based datasets. This leads to more complex grid structures than before for example, in a 3-D x/y/t timeseries data cube, regular lat/long axes of some orthoimagery may be combined with irregular time axis. The OGC unified coverage model allows to describe such generalized grid structures through its concept of a Referenceable Grid Coverage. This standard, utilizing GML, is extended by GML 3.3 with new types of geometries that can handle irregular and warped gridded topologies. A shortcoming is, however, that they are efficient only in the special cases described initially. Hybrid situations like the one above can become inconveniently verbose. In this paper, we document first considerations to improve flexibility and modularity of the GML grid models as part of our work to fully unleash all grid types for OGC type Web services. Indeed, in case of hybrid rectilinear and curvilinear grid models there is room for optimization in the existing GML types, presented by means of formal structure definitions and practically motivated examples.