Powered by Gro

Gro Provides Accurate Forecasts Months Before Government Reports Are Released

In forecasting, accuracy is everything, especially when a forecast lands early. With this in mind, Gro Intelligence has developed a suite of global agricultural analytics and models that accurately predict crop yields and acreage far ahead of official government releases. 

Gro’s suite of machine-learning models currently cover 80% of global production for corn, soy, and wheat. 

Typically, Gro’s US crop yield and acreage predictions are accurate four to six months ahead of official government estimates. In other critical regions, including Russia, Brazil and China, Gro’s forecasts are accurate up to 12-24 months before official estimates are released. Our forecasts are within 2-5% of final government estimates. 

In the US, for example, Gro’s corn and soybean yield models correctly anticipated directional changes ahead of the USDA. Twice this fall the USDA adjusted its yield forecasts to align with Gro’s forecasts for corn and soybeans, as we reported here and here.

We can do this because we have organized, structured, and visualized the world's climate and agricultural data into searchable and accessible information that updates daily. 

Our yield forecast models leverage pixel-level satellite data that capture every field in every county of a state, not just random fields as with traditional crop survey methods. Our yield model forecasts are significantly more accurate because:

  • Research and data science - Our team of experienced data scientists work closely with our research analysts to develop the best possible models. We use human intelligence which is then scaled through artificial intelligence.
  • Comprehensive global data sources and variables - Our forecasts incorporate a combination of Gro’s extensive suite of climate, environment, and crop condition data alongside and variables.
  • Geospatial expertise - Gro’s geospatial team creates proprietary crop masks for each yield model to enhance the explanatory power of climate variables.
  • Machine and human learning - The machine-learning models constantly improve as additional ground truth data comes in, and Gro’s domain experts examine this data to create new predictive features.

Yield Forecast Models: Accurate Well Before Government Estimates

Gro’s Yield Forecast Models use spatially explicit environmental data and ground truth data to monitor crop conditions in-season and to recalibrate final yield forecasts at district, province, and/or national levels every day. They cover corn, soy, and wheat for the US, Argentina, Australia, Brazil, Canada, China, India, Russia, and Ukraine.

The models enable market participants to gauge crop availability and crop prices, monitor crop production, and make early-season forecasts of end-of-season yield. 

Gro yield projections are available months before final government estimates. Our Yield Forecast Models update daily and accuracy increases as the season progresses.

Get certainty on Argentina Soybean Yields five to six months before the Ministry of Agriculture’s (MAGyP) final number is released.
Get certainty on Russia Winter Wheat Yields eight months in advance of the final number from EMISS (Russian Official Estimates).

Accurate Acreage Forecasts Earlier 

In the ag space, government planting intention and prevent plant estimates are helpful for predicting available crop supply for the coming season. However, earlier access to accurate acreage forecasts can help market participants and government agents pursue opportunities and foresee risks ahead of competitors. 

Using Gro’s Planting Intentions and Prevent Plant Models, market participants can get planted area estimates up to six months before official government estimates are released. 

While our Planting Intentions Model considers a complex set of price relationships to predict farmer behavior, our Prevent Plant Model uses spatially explicit environmental data to predict the amount of acreage the farmers will be unable to plant. 

Both machine-learning based models forecasts are within 2-5% of final government estimates. 

Get an accurate forecast 10 months before the final NASS report.

Integrated Analysis for the Ag Sector

At Gro, we have over two million unique agricultural analytics that address a range of questions across yield and production, supply and demand, growing conditions, and climate scenarios. 

Because our models cover a wide variety of subjects, geographies, and environmental conditions, market participants can use them to spot trends and to explore the intersectional effects of supply, demand, price, climate, pests, and disease on crops and on agricultural commodity trade flows. 

Our mission is to help users discover meaningful actionable insights and to facilitate more informed and faster decision making. 

Schedule a demonstration of our platform to learn more about our analytics and forecasts.

Read the full report on the performance of our Yield Forecast Models and Planted Acreage Models.

Get a demo of Gro
Talk to our enterprise sales team or walk through our platform