7 March 2019

The Gro Argentina Soybean Yield Model has maintained its estimate of 3.15 tonnes per hectare, which implies production of 55 million tonnes, on par with the USDA’s February estimate. Significant NDVI and lower ET (evapotranspiration) values have a strong effect on our yield estimates.  

About the Argentina Soy Model

The corn modeling experience created a framework with which we have moved on to further estimation tasks such as soybean yields in Argentina. Data quality and quantity vary around the world, and we had to make adjustments for that. The fundamental concept behind the soy model is similar to the corn—we used district level data and aggregated up to the national level. The model is also similarly updated daily for Gro subscribers and weekly on our website during the growing season. We used many of the same variables, such as:

  • Normalized difference vegetation index (NDVI)
  • Land surface temperature (LST)
  • Soil properties
  • Cropland Masks

The main differences between the models are the how the variables were transformed within the model. For example, for cropland masks, we needed to obtain the data very differently for Argentina soy than for US corn. For the US, we used annual masks from CDL for all historical years, and for the current year, we created a static mask based on historical masks. Argentina does not have an equivalent crop mask source, so we created our own soybean masks from 2001-2016 by classifying pixels.

The machine learning models selected are also slightly different for the Argentina soy model. The US corn yield model uses XGBoost and Cubist first, and then runs it through an XGBoost metamodel. The Argentina soy model only uses a single XGBoost model because the number of training instances is much lower. The metamodel was also unnecessary because the years back tested did not include severely anomalous years. The metamodel therefore did not show different results.

Similar to the US corn yield forecast, you can view the daily forecast on the Gro data platform.

Download our US corn yield model paper
Contact sales