Gro’s Yield Forecast Models use a suite of machine-learning models to estimate in-season yields at the district, province, and/or national levels on a daily basis. Gro’s China Corn Yield Model employs three separate models to track one public and two private data sources. The model contains province-level ground truth data going back to 2010.
Customers Use the Model to:
- Predict province-level yields for the seven main corn-producing provinces in China, which together account for over 70% of China’s corn production
- Make data-driven business decisions on issues that are dependent on Chinese corn production
- Monitor crop production, the most significant part of the balance sheet, by tracking its most variable component, which is in-season yield
- Gauge crop availability and crop prices
- Understand how weather impacts yields in microclimates
- Inform other models focused on damage caused by pests and diseases
Why It Matters
China is a major player in crop production and consumption patterns driving global trends and dynamics. Domestic Chinese corn consumption has more than doubled in the past decade, largely due to demand for livestock feed. Currently the country consumes over 280 million tonnes of corn annually, meaning that Chinese corn inventories have significant influence on the global corn balance sheet. However, it can be challenging to obtain public sources for in-season yield estimates for China. This model fills the gap, enabling users to keep track of yield figures in the main corn-producing provinces.
These machine-learning models are driven by inputs reflecting long-term trends as well as in-season changes. The models run daily throughout the growing season, and rely on province-by-province signals derived using our domain expertise. They rely on proprietary land cover data, built using high resolution satellite imagery to match estimates of area planted from each source. Major model inputs and features include:
- Normalized Difference Vegetation Index (NDVI) signals of corn fields, based on Gro’s specific land cover maps. Satellite-derived NDVI is a powerful indicator of plant health that uses infrared light waves to detect positive signals and plant stressors.
- Temperature from Global Historical Climatology Network (GHCN)
- Soil texture