HydroKG – A national hydrological knowledge graph for Australia 

The Challenge  

Access to hydrological data in Australia is limited in the way it is published, made available and managed. These include the hydrological observation data and reference geographic features – e.g. monitoring stations, gauges, river networks, catchment boundaries, and related observations. Data exists as static files and often requires GIS expertise to extract the required information for use in research projects. The problem with this is that this presents friction across the sourcing, use and reuse of fundamental hydrological data. Furthermore, data access is siloed and is easily integrated across the hydrological domain and across domains. 

Our response

The opportunity is to develop streamlined access to the hydrological reference geographic features and observation data and integrate them in a way that allows inter-related information to be easily queried. This is achieved via a knowledge graph of hydrological features, that is, the HydroKG that enables integrated querying of observations. A knowledge graph of hydrological features provides the conceptual backbone for describing and relating fundamental hydrological concepts, such as, representing water bodies and their narrower types (like catchments, lakes) and river networks. HydroKG will seek to provide this conceptual backbone for all hydrological features in Australia using readily available datasets such as the GeoFabric and HydroATLAS. The conceptual backbone that is provided by HydroKG allows observational data to be associated to the respective real world features and query capability in a rich and precise way, e.g. for the River Murray (real world feature concept), show me all available water quality observations between 1900-2022 (query).  

We propose actively pursuing and building collaboration linkages with partners internationally via the Open Geospatial Consortium and other Knowledge Graph initiatives. This will provide HydroKG with the opportunity to leverage standards established in those forums and influence future design.  

Results

In the 2021 Seed Funded round, a HydroKG pilot project ran from September-December 2021. The project goals were to scope technical and science feasibility, identify key stakeholders within CSIRO and external collaboration linkages, and develop demonstrators. A key output was to draft a system design for HydroKG and assess the maturity of the components needed. In developing a system design, the team implemented a small prototype using the Stream Network dataset from the Bureau of Meteorology’s Geofabric v3.2. The aim of the prototype was to demonstrate an end-to-end platform using a constrained set of real-world features represented in HydroKG (i.e. all stream networks and rivers in Australia), and explore ideas of what could be possible by having these features available not only as a database, but via web APIs. The team developed the Australian River Runner application using these prototype HydroKG APIs, which allows users to simulate the journey of a drop of water from anywhere in Australia running down to the nearest stream and down the river to the final destination outlet.  In doing so, we demonstrated a novel interactive and engaging visualisation capability which relied on the HydroKG APIs. This capability has sparked a number of conversations within CSIRO about integrating more features within hydrology (e.g. catchments, aquifers) and across domains (e.g. environmental accounting, bushfire and landscape modelling) to explore future data analytics capability.  

If you would like to see the demonstration video click the link here HydroKG Demonstration.

For more information contact Jonathan Yu email:jonathan.yu@csiro.au