The 5th WaterSENSE project newsletter is now available

HydroNET Australia and Water Technology are very happy to announce the 5th Newsletter for the WaterSENSE project, a four year research in action project funded by the European Commission.

The goal of WaterSENSE is to develop a modular, operational, water-monitoring system built on Copernicus EO data and provide a toolbox to make products and services available to users.

Click here to read the newsletter.

Newsletter Highlights:


  • WaterSENSE Consortium Visit to Australia.

    • After a long wait, representatives of the European partners from the WaterSENSE consortium visited Australia.
  • Latest Research Feedback:

    • Updated Workflow for ETLook and HSP Algorithms.

      • The previous workflow relied heavily on static third-party land cover maps, which hampered coverage and scalability. For example, irrigation at sites not in the land cover map were missed. In the new workflow, all agricultural classes are now processed, resulting in a ten times larger potential delivery area. Further to this, efficiency improvement have been made. Natural pixels are no longer compared at their native resolution, but are now segmented into blocks of multiple pixels. This segmentation did not only increase the computational performance of the model, but also enabled a drastic increase in the search area, therefore making the model more stable and less vulnerable to local extremes.
Improved Irrigated water use detection coverage.

Significantly larger area for which irrigated water use can be estimated in the new ETLook and HSP workflow.


    • Automated Irrigated Area Detection.

      • eLEAF has started testing its capabilities for detecting irrigated areas. Some simple statistical metrics were calculated based upon the irrigated water use estimates (ETinc) from eLEAF’s HSP algorithm. With these metrics, a simple random forest model was trained. The feature importance of each of these metrics was then evaluated and finally, the trained model was used to compute annual irrigated area maps for different years (2017-2020). The standard deviation and maximum of ETinc explained the majority of the outputs, indicating that ETinc data alone is able to accurately estimate the irrigated area without adding any auxiliary dataset, such as the more general satellite based indices like NDVI and NDWI.
Simple Random Forest Model to detect Irrigated area with ETinc metrics.

Training samples vs ETinc. ETinc is able to discern irrigated training samples from rainfed training samples across all sites.


    • Open Water Detection and “growing” detected pixels under the trees.

      • An extension of the open water detection algorithm for wetlands and riverine areas is currently being developed. In particular, the algorithm development focuses on solving identified problems with clouds, vegetation and coloured water interference in the open water detection as well as the detection of surface water under vegetation. The novel geomorphic growing algorithm uses the digital elevation model to “grow” satellite-identified water patches to be hydrologically sensible and connected (See figure below). This work was recently presented by the USYD WaterSENSE partner at the international Environmental Modelling and Software Society meeting in Brussels (July 2022). This algorithm will be further verified and completed by December 2022.
Water detection and pixel "Growth" to fill in gaps under trees.

WaterSENSE’s satellite-identified water detection algorithm output in yellow. Geomorphic approach to “grow” these pixels to hydrologically sensible and connected areas under the trees in violet.


    • Vegetation Condition Parameters.

      • WaterSENSE aims to provide timely information on the extent, health and water-use of wetland vegetation from remote sensing data and algorithms. One of the parameters we are exploring is the Biomass Production Condition (BPC) Score. This indicator tells us how the biomass production of an individual pixel is changing over time. The figure below shows some initial results in the Barmah Forest, Australia. It highlights how rapidly vegetation condition has changed from very high in October 2017 (in green) to very low in October 2021 (in red).
WaterSENSE Biomass Condition

Biomass condition maps and graph show the deterioration of the vegetation condition from 2017 to 2021. Legend Explanation: Biomass condition represents a measure of plant production at a scale from 0 ‘the lowest’ to 1.0 ‘the highest’ in relation to reference (optimal) plant production


  • 2nd Summer School Outcomes –  23 to 25 February 2022.

  • List of Publications.