An AI-powered forecasting firm currently expects China, India and the European Union to underachieve USDA forecasts for wheat, while Argentina, Brazil and China are seen missing the mark on corn production. The firm also sees slightly lower production of soybeans in the South American powerhouses of Argentina and Brazil.
The Australia-based company, Octopusbot, sees the U.S. on track to overshoot on both corn and wheat.
Octopusbot utilizes a variety of weather data including traditional variables like temperature and rainfall, but also factors in other advanced information like solar radiation, wind speeds, and even snow depth for winter crops. Additional information from satellite data, such as NDVI (Normalized Difference Vegetation Index), is also taken into consideration. The data is input into a regression model that works to predict yields based from the multitude of variables and a machine learning algorithm.
Pro Farmer has tapped Octopusbot to provide high-frequency weather and satellite data from around the world. Data from Octopusbot is a useful tool for monitoring conditions throughout the year, but it should also be emphasized that our Pro Farmer Crop Tour estimates will be the primary indicator that we look at when providing our annual yield estimates. Objective, on the ground data that is sampled shortly before harvest time remains the pinnacle of yield predictions in agriculture.
Meanwhile, weather concerns have returned to the grain markets. Heatwaves have gripped most of western and central Europe this summer while other top wheat producing regions have received drought relief. In our own hemisphere, corn pollination is in full swing in the Midwest, while El Nino looks to potentially curb new-crop production potential in South America.
USDA’s Interagency Commodity Estimates Committee, the group that prepares the WASDE, also examines weather data when setting their own estimates. However, that process is also influenced by attaché reports provided by employees in the Foreign Agricultural Service, as well as data provided by other foreign governments, though USDA notes that those reports are still scrutinized. Octopusbot does not factor in these, preferring to rely solely on their regression and machine learning models. USDA provides useful data services that are also qualified forecasts, though can have a tendency to adjust yields in foreign markets at a slower pace to avoid making a move too early or stray too far from another government’s official estimate. The example of Argentinian corn production, where analysts had pegged production at its current level months ahead of USDA making the adjustment is one instance from recent memory.
As of July, Octopusbot sees wheat and soybean production in top producing countries as lower than USDA, while corn production is forecast to be higher. As the table below shows, the firm expects production of wheat to be much lower in China and the EU, with too much rain occurring at poor times in Asia, and too little rain during the grain fill period in Europe. More favorable weather in the Black Sea and Australia in recent months has helped to offset most of the losses elsewhere, however.
Corn production is seen as lower across most of the globe in the major production regions, though the U.S. estimate is notably higher, as the Octopusbot model falls firmly in the grain makes rain school of thought. The forecaster currently sees the rainfall that occurred between April and May, which was 128% of average, and the July to October precipitation outlook that also calls for higher than average rainfall as significant yield boosters. The current forecast is 19.7 MT, or 4.8%, higher than USDA’s production outlook. If disease pressure ramps up as the summer progresses, that could quickly turn the corn market on its head as production is already expected to be somewhat lower in other parts of the world. Variables such as NDVI and rainfall can provide clues to canopy growth and other factors, but can sometimes miss potential issues that are more readily evident on in-person analysis like disease pressure or pollination issues.
Time remains in the growing season, and weather will always be a fluid variable that is hard to perfectly forecast. While no two growing seasons are the exact same, taking an objective look at how weather conditions have impacted previous years can provide valuable insight to help us prepare and adjust marketing plans based on expected changes.