Graincast™: forecasting Australia’s national grain crop yield

Graincast™: the digital infrastructure that maps, monitors, manages and forecasts the Australian grain industry’s production across vast areas of our continent.

What problem does Graincast™ solve?

Grain production is closely tied to climate and is one of the most volatile industries in Australia, making planning and managing a huge challenge.

Unfortunately, the information base has been limited. Planning for harvest, transport, storage and marketing is based on historical records supplemented by subjective assessment of current crop conditions. There is no comprehensive national system for quantifying and geo-locating crop area, and the size of the national grain crop area and expected yield is never known with required accuracy at a locally relevant scale.

Nevertheless, there has been some uptake of decision tools such as Yield Prophet® and iPaddock, showing that there is appetite for services that supply data, provided the output is of value and packaged in a user-friendly manner.

Each actor along the grains supply chain has a unique need for customised grains information, which could be met with a comprehensive grains production and forecasting information system. It needs to provide up-to-date estimates of crop area, near real-time measures of crop yield, crop yield forecasts, and estimates of uncertainty for the entire agriculture value chain.

What could Graincast™ be used for?

This is where Digiscape came in. With CSIRO’s expertise in grain forecasting, climate, remote sensing, data management, software engineering, social innovation and more, we built a real-time grain forecast platform for Australia: Graincast™. It could open up greater informed decision-making, such as:

• farmers being able to predict yield within-season to inform input management and forward-selling
• consultants able to advise farmer clients on a portfolio of cropping enterprises to manage risk and return
• bulk handlers having more precise estimates of crop area and yield down to individual farm scale, allowing better planning and utilisation of transport and storage
• grain marketers tracking and monitoring the global grain supply better and taking advantage of market fluctuations.

The Graincast™ forecast system could also open up the capability for new decisions, such as:

• farmers (individually or aggregated) able to seek better deals on inputs from resellers
• resellers being able to supply more product where and when it is needed
• bankers able to make more transparent lending decisions for seasonal or capital finance
• insurers able to objectively assess risk profiles before setting premiums for products such as multi-peril crop insurance.

How do I access Graincast™?

In early 2020, we licensed the exclusive global rights to the Graincast™ smarts behind the app to Australian rural technology start-up Digital Agriculture Services (DAS). For the first time, farmers can see their 2019 Graincast™ results integrated with a range of other, publicly available agri, rural and climate risk insights – at no cost – on DAS’ Rural Intelligence Platform™.  

Thus, we are no longer providing Graincast™ as an app but we hope it helped you make better on-farm decisions while it was available. Graincast™ provided paddock-level yield forecasts for over 400 farmers, agronomists and sticky beaks during its four years of operation. And it won a state-level award at the Australian Information Industry Association’s iAwards competition.

The Graincast™ app enabled the Australian grain growing industry to utilise breakthrough Australian science, and now it’s evolved into something even bigger.

CSIRO’s national wheat yield free, fortnightly forecasts: Wheatcast™

However, we’re still providing select forecasts of Australia’s wheat yield each fortnight during the grain growing season, at no cost. But, we’ve changed their name and we’re now calling these forecasts Wheatcast™.

In 2021 we added some bells and whistles with maps of soil water, water-limited yield forecasts and their uncertainty, and state specific forecasts. Wheatcast’s™ forecasting capability uses our innovative analytics and is an Australian first.

Daily rainfall, temperature and solar radiation data are sourced from the Bureau of Meteorology at 202 selected high quality observation stations. Soil data are sourced from the Australian Soil and Landscape Grid and matched to these weather stations. Soil water status and water-limited grain yield forecasts are calculated by the Agricultural Production Systems Simulator, APSIM®, a modular modelling framework developed to simulate biophysical process in farming systems. APSIM® contains a suite of modules that include a diverse range of crops and pastures, and soil processes including water balance and nitrogen transformations. We also use statistical analysis of past annual yields data to convert water-limited yield potential into actual yields achieved at national and state levels.

See the fortnightly wheat yield forecasts.

Where can I find more information on Graincast™?

• visit the Graincast™ website
• see the Wheatcast™ fortnightly wheat yield forecasts
• read our media releases: Melbourne tech company taking crop forecasting global (6 Mar 2020) and New app gives farmers real grain of truth (28 Jun 2018)
• Watch team member Randall Donohue present the science behind the C-Crop yield model used by Graincast™:

• Read select publications:

Lawes R, Hochman Z, Jakku E, Butler R, Chai J, Waldner F, Donohue R (2022) Graincast™: monitoring crop production across the Australian grainbelt. Crop & Pasture Science. doi: https://doi.org/10.1071/CP21386

Chen Y, Donohue RJ, McVicar TR, Waldner F, Mata G, Ota N, Houshmandfar A, Dayal K, Lawes RA (2020) Nation-wide crop yield estimation based on photosynthesis and meteorological stress indices. Agricultural and Forest Meteorology 284, 107872. doi:10.1016/j.agrformet.2019.107872

Chen Y, McVicar TR, Donohue RJ, Garg N, Waldner F, Ota N, Li L, Lawes RA (2020) To blend or not to blend? A framework for nationwide Landsat-MODIS data selection for crop yield prediction. Remote Sensing 12, 1653. doi:10.3390/rs12101653

Dayal K, Brown JN, Waldner F, Lawes RA, Hochman Z, Horan H, Donohue RJ, Chen Y (2020) Climate drivers provide valuable insights into late season prediction of Australian wheat yield. Agricultural and Forest Meteorology 295, 108202. doi:10.1016/j.agrformet.2020.108202

Fowler J, Waldner F, Hochman Z (2020) All pixels are useful, but some are more useful: Efficient in situ data collection for crop-type mapping using sequential exploration methods. International Journal of Applied Earth Observation and Geoinformation 91, 102114. doi:10.1016/j.jag.2020.102114

Kamir E, Waldner F, Hochman Z (2020) Estimating wheat yields in Australia using climate records, satellite image time series and machine learning methods. ISPRS Journal of Photogrammetry and Remote Sensing 160, 124-135. doi:10.1016/j.isprsjprs.2019.11.008

Radoux J, Waldner F, Bogaert P (2020) How response designs and class proportions affect the accuracy of validation data. Remote Sensing 12, 257. doi:10.3390/rs12020257

Waldner F (2020) The T index: measuring the reliability of accuracy estimates obtained from non-probability samples. Remote Sensing 12, 2483. doi:10.3390/rs12152483

Zhao LY, Waldner F, Scarth P, Mack B, Hochman Z (2020) Combining fractional cover images with one-class classifiers enables near real-time monitoring of fallows in the northern grains region of Australia. Remote Sensing 12, 1337. doi:10.3390/rs12081337

Fritz S, See L, Laso Bayas JC, Waldner F, Jacques D, Becker-Reshef I, Whitcraft A, Baruth B, Bonifacio R, Crutchfield J, Rembold F, Rojas O, Schucknecht A, Van der Velde M, Verdion J, Wu BF, Yan NN, You LZ, Gilliams S, Mücher S, Tetrault R, Moorthy I, McCallum I (2019) A comparison of global agricultural monitoring systems and current gaps. Agricultural Systems 168, 258-272. doi:10.1016/j.agsy.2018.05.010

Waldner F, Bellemans N, Hochman Z, Newby T, de Abelleyra D, Verón SR, Bartalev S, Lavreniuk M, Kussul N, LeMaire G, Simoes M, Skakun S, Defourny P (2019) Roadside collection of training data for cropland mapping is viable when environmental and management gradients are surveyed. International Journal of Applied Earth Observation and Geoinformation 80, 82-93. doi:10.1016/j.jag.2019.01.002