Gro Intelligence – An AI company that improves decision-making in the areas of climate, food and agriculture

Gro Intelligence has added an Australian wheat yield model to its suite of high protein wheat yield forecasting tools to monitor global wheat harvests.

Using Gro’s Australian wheat yield model model, standard users can get an early reading of Australia’s high protein wheat yields at the sub-state level. For example, confirmation of a bumper crop in Australia can cushion crop losses elsewhere and allow users to make better purchasing decisions.

Australia is expected to harvest a bumper crop of high protein wheat from October 2021. This could help offset disappointing high protein spring wheat harvests in Canada, the United States and Russia.

Recently, the Australian Bureau of Economics and Agricultural Sciences and Resources increased its estimate of national wheat production to 32.6 million tonnes, 17% higher than its estimate in June but below harvest. exceptional from last year. Typically, Australia produces 17 to 33 million tonnes of wheat per year; yields often vary considerably in drought years.

As Australian white and amber wheat accounts for 11.5% of global exports and an even larger share of high protein wheat exports, its wheat plays an important role in the global wheat supply.

This means that having an overview of Australian wheat harvest conditions is essential for establishing a global balance of wheat supply and demand and for understanding how trade flows will react. With this in mind, we are integrating our new Australian wheat yield model into Gro’s automated balance sheets, which are used by market players in purchasing, sovereign food security, shipping and asset management to manage supply risk, forecast prices and forecast trade flows.

Today Australia is the closest producer to the Asian market for protein-rich white wheat, a type of wheat commonly used for making noodles. Australian wheat is also gaining more and more followers among food and beverage buyers in China.

Our new Australian wheat yield model uses the Australian Bureau of Statistics (ABS) as the source of ground-verified data. And like Gro’s other wheat yield forecasting tools, our Australian wheat yield model:

  • Daily updates

  • Factors related to meteorological and climatic variables

  • Uses a crop mask based on satellite imagery

  • Descent to sub-state level

With a mean absolute percent error (MAPE) of 4.5% (as of November 1), the yield estimate from the Gro wheat yield model in Australia was significantly more accurate than the USDA estimate, which had an EAP of 7.0% (WASDE report from November).

The accuracy of Gro’s Australian wheat yield model aligns with other Gro wheat yield models, despite a limited history of sub-state ABS data, which goes back 10 years. To address this mismatch, the Australian wheat yield model’s machine learning algorithm, XGBoost, tests more data points and improves its accuracy as it goes. The inputs to our Australian wheat yield model include:

  • Normalized Differential Vegetation Index (NDVI)

  • Gro Drought Index (GDI)

  • Precipitation (NASA GPM)

  • Evapotranspiration anomaly (ETA)

  • Earth’s surface temperature (LST)

  • Soil moisture (SMOS)

Gro’s wheat yield models cover over 50% of global wheat production. Our other wheat models include:

To learn more about the Australian Wheat Yield Model or any of our other forecasting models, contact our sales team [email protected] for a demonstration of the Gro platform.

About Lolita Plowman

Check Also

Burning questions about Thanksgiving turkey

We’ve all seen stories of how rising inflation drives up the price of Thanksgiving dinner …

Leave a Reply

Your email address will not be published. Required fields are marked *