Information about localized phenomena may be represented in multiple ways. GIS systems may be used to record the spatial extent of the phenomena. Observations about the state of one or more properties of the phenomena are available from real-time sensors, models, or from archives. The relationships between these data sources, or specific features in different data products, cannot easily be specified. Additionally, features change over time, their representations use different spatial scales and different aspects of them are of concern to different stakeholders. This greatly increases the number of potential relationships between features. Thus, for a given feature we can expect that heterogeneous information systems will exist, holding different types of data related to that feature. We propose the use of Linked Data to describe the relationships between them. We demonstrate this in practice using the Australian Hydrologic Geospatial Fabric (Geofabric) feature dataset and observational data of varying forms, including time-series and discrete measurements. We describe how different resources, and different aspects and versions of them, can be discovered and accessed. A web client is described that can navigate between related resources, including using the Geofabric’s feature relationships, to navigate from one observational dataset to another related by hydrological connectivity.