Smart water use detection: AI-powered parcel-level water use monitoring for the Darwin rural area
The Water Resources Division has developed an internal web-based analytical tool to estimate vegetation water use on land parcels.
By integrating satellite data, climate archives, and AI-driven analytics we identify properties where water is used by vegetation and may exceed sustainable limits.
The application identifies total water use by vegetation, which is a combination of all transpiration loss from native vegetation, irrigated pastures and commercial crops.
Unregulated groundwater extraction poses a risk to water users and dependent ecosystems. Current commercial water use is monitored through reports by licence holders in accordance with water licence conditions.
The volume of unlicensed water use for stock and domestic use is based on estimates from limited metering and historic investigations. Unlicensed commercial water use can lead to drawdown within aquifers and interference with adjacent water bores. This new application provides insights into water use that can guide compliance efforts and ensure sustainable water management.
Methodology and data sources
This system uses high-resolution satellite data, which can track water usage with a high degree of accuracy, capturing small variations in plant water consumption and soil moisture across a property. The system examines water data from pixels that cover an area of 30 by 30 meters and then combines them to calculate the total water use.
The high resolution allows detection of small-scale water use, such as backyard irrigation or localised groundwater pumping. This makes it an effective tool for monitoring compliance, managing water resources sustainably, and identifying areas where water use activities may need field verification and inspection.
This tool enhances the department's ability to conduct targeted compliance activities by identifying high-risk properties for field inspections or water meter audits. It also allows the team to track seasonal or long-term usage patterns and detect anomalies. The figure below shows an example of monthly analysis of vegetal water usage data for an investigation parcel.
Future enhancements
Future enhancements include integrating machine learning for automated anomaly detection, enabling the system to flag sudden usage spikes in real time. The team also aims to expand coverage to other high-priority regions, such as Katherine and Mataranka. Future iterations may incorporate licensing data, cross-referencing abstraction permits with inferred usage to improve compliance accuracy.
Conclusion
This new application represents a significant advancement in evidence-based water regulation, enabling proactive management of groundwater resources. By combining remote sensing with scalable analytics, the department can monitor water use more efficiently, identifying areas of water use which exceed licensed entitlements or estimated volumes in the case of unlicenced water use.
