Future-Focussed Space and Spatial Data
Written by Kate Williams (FrontierSI) and Prof Matt Duckham (RMIT University)
Purpose
This article provides a short introduction on how advanced spatial data modelling can harness the rich information derived from the ever-increasing stream of temporal data from space-based sources and create valuable, new, insight-ready products and services.
Introduction
For over 30 years, we have been capturing imagery of the earth from space, with new sensors and products proliferating, and almost too overwhelming to keep up-to-date with! The volume and variety of data available increases every day, and the challenge of the geospatial industry has been to remove the barriers to use by creating standardised datasets that can support a wide range of purposes and end-users.
The solution has often been to create vector-based thematic datasets that represent a single authoritative geometry of a feature, for example, a polygon dataset of waterbodies that represents the average extent of lakes, dams, wetlands, and so on. If the area is prone to drought or flood, there could be more individual datasets created to represent these different geometries from different dates. However, this model of layered data storage doesn’t fully capture and unlock the value of temporal space-based data to create new insights, particularly for those who are not skilled spatial analysts.
To truly capture the value of space-based data streams, we need to rethink the “layer cake” model of disparate spatial datasets and move to modern integrated data models that link multiple representations and features that are being constantly enriched and improved.
Opportunity for change
As mentioned, the volume of space-based imagery increases every day, but this volume is no longer a barrier for analysis, with computing innovations including cloud computing and perhaps quantum, unlocking this big data and removing the constraints on storage and analysis. There are some fantastic examples of the creation of analysis-ready imagery infrastructure utilising Open Data Cube technology, such as Digital Earth Australia, that enables multi-temporal analysis of huge imagery databases, and these could be the foundation of moving geospatial data from being raster “Analysis Ready” to vector-based “Insight Ready” data utilising knowledge graphs and ontologies.
This innovation will facilitate a wider range of more complex queries for expert and non-expert users. It will facilitate the use of machine learning (ML) and Artificial Intelligence (AI) to leverage the potential of chat-based queries for spatial data, generating insights for a wider variety of users using simple plain language questions. For example, if a knowledge graph database holding multiple representations of lakes existed, a user could ask “When was the lake surface area the smallest? Or how many lakes reduced in size by over 10% in the last 5 years?”. Using the traditional layered dataset storage, these queries would involve multiple different datasets and advanced spatial analysis.
This envisioned future will truly revolutionise the way we unlock the value of big data, including space-based data, and create richer data, to enable better representation of our dynamic world and enhance our decision-making through new insights.
What next?
This short article has used the example of water bodies to spark your interest in moving from analysis-ready geospatial data to insight-enabling geospatial data using space-based data streams. But this is only the starting point of the opportunity!
Knowledge graphs and ontologies facilitate the breaking down of data silos across domains, and truly allow us to create a digital twin of the world, incorporating the way systems and features relate to each other. Ontologies facilitate the representation of how a lake links to the river that feeds it, to the catchment it is in, to the land use, and land use change that impacts water quality, and more.
For more information on a case study on how Knowledge Graphs and Ontologies were demonstrated in Victoria, please see the following resources:
Application Areas | Projects (frontiersi.com.au)
Enriching spatial data through Dynamic Vicmap project (land.vic.gov.au)
Thanks to the SmartSat CRC and the Victorian Department of Transport and Planning for funding the work that led to this short article, and to RMIT for generously sharing your knowledge on this topic.
