From surveys to satellites: a new era for irrigation water estimates

Water is one of Australia’s most valuable resources, underpinning the economy, ecosystems, and communities. Leveraging the expertise of the Commonwealth Scientific and Industrial Research Organisation (CSIRO), the Australian Bureau of Statistics (ABS) has modernised the measurement of agricultural irrigation water use through the adoption of a satellite-based evapotranspiration method, developed by CSIRO with support from the Terrestrial Ecosystem Research Network (TERN). This innovation uses satellite imagery and climate data to estimate irrigation water use by agriculture across Australia - delivering nationally consistent, timely, and sustainable data without burdening farmers. Learn more about the science behind these estimates at TERN’s evapotranspiration data portal.

For decades, ABS measured agricultural irrigation water use through its Agricultural Survey and Census. These surveys provided critical data for the Water Account, Australia, helping governments, businesses, and communities understand how water was used across the country.

When these surveys ended after 2021–22, a major data gap emerged. Irrigation is a key component of agricultural water use, and without reliable data, the Water Account risked losing one of its most important indicators. How could we continue to track irrigation without imposing new burdens on farmers?

The solution lies in the application of satellite technology. Beginning with the 11 December 2025 release of Water Account, Australia, these CSIRO-developed estimates will be published as experimental, marking a major milestone in modernising water statistics.

For nearly two decades, ABS relied on survey responses to estimate irrigation water use by agriculture. While robust, this approach had limitations:

• Respondent burden: farmers faced lengthy questionnaires, adding to their workload.
• Coverage gaps: only businesses with an Estimated Value of Agricultural Operations (EVAO) over $40,000 were included, excluding smaller irrigators.
• Declining response rates: rising costs and survey fatigue reduced data quality.

The discontinuation of the Agricultural Survey/Census aligned with the ABS Agriculture Statistics Modernisation Program, which seeks innovative, cost-effective data sources. Without a replacement, the Water Account risked losing a key component of agricultural water use. Enter CSIRO’s evapotranspiration model – a method that uses satellite imagery and climate data to estimate agricultural irrigation volumes without a single survey form.

Evapotranspiration (ET) combines two processes:

• Evaporation: water turning into vapour from soil and surfaces.
• Transpiration: plants releasing water through their leaves.

Actual evapotranspiration (AET) measures this under real-world conditions, while potential evapotranspiration (PET) assumes unlimited water availability. By comparing AET with rainfall and soil moisture, scientists can infer how much extra water – irrigation – was applied.

CSIRO’s method uses the CSIRO MODIS Reflectance-based Scaling Evapotranspiration (CMRSET) algorithm to estimate AET across Australia at fine spatial resolution. These data feed into a water balance model that calculates irrigation as the residual water needed to meet crop demand after accounting for rainfall and soil moisture. In simple terms:

Irrigation = Water needed – Water nature provided.

The CSIRO evapotranspiration method integrates satellite observations, climate data, and land-use information to estimate irrigation water use for agriculture across Australia. Here’s how:

1. Mapping irrigable land – using the Australian Bureau of Agricultural and Resource Economics and Sciences’ (ABARES) land-use datasets, the model identifies areas where irrigation is likely, such as irrigated cropping, horticulture, and pasture.
2. Defining irrigation periods – the model uses satellite vegetation data to determine crop growth stages and seasonal patterns, ensuring irrigation timing reflects real-world practices.
3. Compiling water balance inputs
   • AET: estimated using CSIRO’s CMRSET algorithm applied to MODIS satellite imagery.
   • Rainfall and PET: sourced from the Bureau of Meteorology’s Australian Water Resources Assessment Landscape model (AWRA-L) model.
   • Soil moisture and runoff:  also derived from AWRA-L.
4. Calculating irrigation water use – the water balance equation is applied monthly to millions of pixels nationwide, creating a detailed spatial map of irrigation activity.
5. Aggregating results – monthly estimates are summed to annual totals and aggregated to states, catchments, and statistical areas, aligning with ABS reporting requirements.

Refer to the Methods – Agriculture irrigation water use section of the Methodology of the Water Account for more information.

The end of the Agricultural Survey and Census created a major gap in irrigation water use data. This new satellite-based approach closes that gap and ensures continuity for one of the most important indicators in the ABS Water Account. Unlike previous surveys, which excluded smaller farms, the method captures irrigation activity across all irrigated areas, providing a more complete and representative picture of water use in Australia.

By replacing manual surveys with automated satellite and model-based estimates, the ABS can deliver data more quickly and at lower cost, while removing the burden of lengthy questionnaires from farmers. The approach also offers greater detail, producing monthly estimates at 100m resolution that reveal seasonal irrigation patterns and allow flexible aggregation to catchments, irrigation districts, or small statistical areas – supporting planning, policy, and sustainability decisions.

The new model has been tested against historical ABS survey data and shows strong alignment, with differences typically within 5%. Validation studies report R² values up to 0.96 and bias reduced to –4.6%, giving confidence that the approach provides reliable national and state-level estimates. Refer to the Methods – Agriculture irrigation water use section of the Methodology page of the Water Account for more information.

Users should be aware that the introduction of this new method creates a break in the time series for agricultural water use in the ABS Water Account. Movements between 2021–22 and 2022–23 may appear volatile, particularly in the split between self-extracted and distributed water use, rather than in total agricultural water use. These changes reflect the shift from survey-based estimates to satellite-derived modelling rather than real-world changes in irrigation practices. Future work will focus on backcasting and refining the methodology to restore continuity and improve comparability over time.

High-resolution agricultural irrigation data provides the evidence base for effective decision-making, including for water resource management. Governments and water authorities could have the potential to use these estimates to plan allocations in major basins like the Murray–Darling, ensuring water is distributed efficiently and sustainably. The data also supports drought preparedness, helping identify regions at risk and guiding emergency water strategies.

Environmental regulators can monitor ecosystem health and environmental flows, ensuring compliance with conservation targets. Beyond regulation, the method underpins market-based instruments such as the Nature Repair Market, where accurate water use data is essential for valuing ecosystem services.

While the CSIRO evapotranspiration method represents a major advance, several challenges remain:

• Surface water vs groundwater: currently, all irrigation water is treated as surface water, but many regions rely heavily on groundwater.
• Crop-level detail: water demand varies significantly by crop type. Future work will integrate high-resolution remote sensing and classification algorithms to map irrigated areas by crop.
• Small-area precision: enhancements will include dynamic irrigation periods that reflect seasonal variability and climate anomalies, as well as integration of administrative data such as water licences.

These refinements will ensure the method remains fit-for-purpose, supporting not only national accounts but also regional planning, sustainability reporting, and market-based instruments.

The adoption of CSIRO’s evapotranspiration model marks a new phase in Australian water statistics. By harnessing satellite technology and advanced modelling, ABS can deliver nationally consistent, timely, and sustainable agricultural irrigation estimates – without burdening farmers or relying on traditional surveys. This innovation strengthens the Water Account, Australia and aligns with international standards, ensuring Australia remains a leader in environmental-economic accounting.

These experimental estimates will not only strengthen the Water Account, Australia but will also underpin the National Ecosystem Accounts, where they will be used to measure water provisioning services from freshwater ecosystems. Harmonising methods across accounts ensures coherence between physical water flows and ecosystem service flows, supporting integrated environmental-economic reporting.

Australian Bureau of Statistics (ABS) 2023, Modernising ABS Agricultural Statistics, ABS, viewed December 2025. 

Australian Bureau of Statistics (ABS) 2025, National Ecosystem Accounts – Experimental Estimates, ABS, viewed December 2025.

Bureau of Meteorology (BoM) 2023, Australian Water Resources Assessment Landscape Model (AWRA-L) v7: Model Description Report, BoM, viewed December 2025. 

Commonwealth Scientific and Industrial Research Organisation (CSIRO) 2023, CMRSET Algorithm Documentation, CSIRO, viewed December 2025.

Terrestrial Ecosystem Research Network (TERN) 2023, Evapotranspiration Data Portal, TERN, viewed December 2025.