Droughts are one of Australia’s biggest natural hazards. They reduce crop yields, threaten livestock, reduce water storage volumes in dams, damage forests, and increase the risk of bushfires. Often, by the time plants show visible signs of stress, such as yellowing leaves and/or thinning canopies, the damage is already done.
A team of researchers, led by Dejun ‘Jack’ Cai and Tim McVicar (CSIRO Environment Research Unit, Canberra), including four other CSIRO Environment scientists (Tom Van Niel, Randall Donohue, Steve Stewart and Matt Stenson) along with Yuhei Yamamoto and Kazu Ichii (partners from Chiba University, Japan) developed a powerful new approach for early detection of vegetation drought stress, giving farmers, land managers, and decision-makers more time to prepare. Their approach, called the Surface-Air Temperature Difference Anomaly (SATDA), uses geostationary satellite thermal data and ground-based gridded climate data to detect the very first signs of plant stress, well before the stress becomes visible. SATDA is one of several outputs from a multi-year Himawari project that is supported by the NCRIS-funded TERN (Terrestrial Ecosystem Research Network) Landscapes platform which develops best-practice Himawari datasets for enhanced monitoring and understanding of sub-diurnal atmospheric and biophysical processes across Australia.
How does it work?
Healthy plants “sweat” by releasing water through tiny pores (called stomata) in their leaves. This process, called transpiration, keeps them cool, just like how sweating cools us down. But under drought conditions, plants close these pores to save water. This means they stop cooling themselves, and their leaves heat up.
SATDA tracks this effect by comparing:
- Land surface temperature (Ts) – measured from space every 10 minutes by the Himawari-8 geostationary satellite (designed, launched and operated by the Japan Meteorological Agency (JMA)), and
- Air temperature (Ta) – measured at Australian Bureau of Meteorological (BoM) weather stations that are spatially interpolated by the CSIRO scientists to form hourly meteorological grids.
From late morning to mid-afternoon when the difference between the surface and air temperature (i.e., Ts – Ta) suddenly increases, it’s a warning sign that plants are struggling. By tracking anomalous changes in Ts – Ta over time, SATDA can detect stress signals up to 12 weeks before vegetation visibly declines.
A real-world test: the Tinderbox Drought
Researchers put SATDA to the test by examining the 2017–2019 Tinderbox Drought in southeast Australia, one of the most severe droughts in Australia’s recent history. Within this prolonged multi-year dry period, SATDA was able to capture a “flash drought” that rapidly unfolded at weekly timescales between July and September 2019 in northeast NSW. Hidden within the larger drought, this event caused sharp agricultural losses and heightened bushfire risks.
Unlike conventional drought indicators (such as precipitation and soil moisture), SATDA provided earlier and more accurate forecasts of visible vegetation decline (as measured by the Normalised Difference Vegetation Index; NDVI), particularly in the transitional semi-arid and sub-humid regions. The method proved especially effective in forests and woodlands, where deep-rooted trees typically make drought impacts harder to detect using traditional approaches that focus on landscape water availability. These results indicated that by leveraging sub-daily Himawari-8 Ts observations and gridded Ta data, SATDA showed strong potential in forecasting the vegetation response to droughts up to 12 weeks in advance and thus support future drought early warning applications.
Powered by NCI’s Gadi supercomputer
Turning satellite data into a nationwide drought-monitoring system is no small task. The research team relied on the National Computational Infrastructure (NCI) to make it possible via:
- High-performance processing: The analysis spanned more than 1.7 million grid locations over ~2,700 days. Thanks to the Gadi, researchers processed all these data and generated all-Australia SATDA images in just two hours by running tasks in parallel across ~1000 CPU cores.
- Seamless data access: NCI hosts extensive climate and environmental datasets. This gave the research team direct access to BOM rainfall grids and hydrological model outputs, which were crucial for evaluating their SATDA results.
- User-friendly tools: Through NCI’s Australian Research Environment (ARE) platform, researchers used interactive software like RStudio to explore the data and create high-quality figures for international publications.
- International scientific data agreement: JMA and BoM have an international data sharing agreement where Himawari data are provided to Australia via a dedicated internet link from Japan to the BoM for operational use. BoM make these observations available to research agencies like CSIRO, Geoscience Australia (GA) and Australia’s university sector via the NCI. The NCI high-performance computing environment enables researchers to process Himawari data volumes in near real-time and to reprocess the data collection (2015-present) as required.
Without this combination of supercomputing power, curated datasets, analysis platforms, and international scientific data agreement, scaling SATDA across Australia simply would not have been possible. All-Australia daily SATDA grids currently reside on the NCI dt1 project and will also soon be freely available from the TERN website.
Looking ahead: from Australia to the world
While this initial study focused on southeast Australia, the process-based, parsimonious and scalable design of SATDA make it suitable for global implementation. The approach can be applied to other geostationary satellites such as GOES-R (Americas) and Meteosat (Europe and Africa), opening the door to worldwide vegetation drought early-warning systems.
In the future, SATDA could form part of an international drought early-warning network, giving governments, farmers, and communities the ability to respond to droughts before irreversible damage occurs. Such early-warning systems can also inform agricultural planning, water supply, disaster preparedness, and climate policy.
This breakthrough has been made possible by combining innovative science with the high-performance computing power of NCI. Using satellite observations, process-based insights, and advanced supercomputing, researchers have created a tool that not only tracks vegetation health with unprecedented precision but also scales seamlessly across all of Australia’s landscapes.
Together, these advances represent a major step forward in monitoring plant health, safeguarding ecosystems, and strengthening climate resilience in a warming, drying world.
