In the United States, agricultural professionals use yield prediction maps for the management of agricultural stocks and increasingly for the assessment of risks for the insurance sector. The United States Department of Agriculture (USDA) publishes monthly maps with yield and area sown per county [1] for the 16 main crops. These maps are either historical and based on observations (from past cropping years) or predictive, and are updated every month for the current cropping year.
However, much of the USDA data comes from expensive and time-consuming surveys. And they are only partially representative since these data are produced from a statistical method from a sample of farmers.
Last but not least, the USDA forecasts do not take into account potential future climatic disturbances (for example episodes of drought or future cold spells). In order to overcome all these limitations, itk has managed to generate yield forecast maps for corn and soybeans, based on plant and crop condition modeling at the county level.
Itk evaluates yield predictions earlier in the season
Faster and less expensive, plant modeling allows yield maps to be obtained by combining weather trends over several months with data for average cultivation practices by county. Generating these maps is possible early in the growing season (well before USDA maps), and gives farmers more time to anticipate and make the best decisions to maximize their yield.
For example, for the year 2018, in South Dakota and Nebraska, the itk model estimated an average yield of 166 bu/ac for South Dakota and the USDA estimated 160 bu/ac.
For Nebraska, the ITK model estimated an average yield of 191 bu/ac and the USDA estimated 192 bu/ac.
Historically, itk has used its agro-environmental models to make forecasts at the paddock level. A change of scale from the plot to the county was necessary and for this, data from satellite imaging techniques were used in order to quantify the cultivated area by species and by county. In corn crops, the first feasibility study gave rise to initial results. It was presented to and validated for our client Winfield, a subsidiary of the Land O’Lakes cooperative.
Cover yield losses due to climatic hazards
The relevance of itk’s forecasts at this territorial scale could open up new opportunities. It can also be an opening to forecasting for other types of crops, as for other sectors of activity.
The agricultural insurance industry is particularly interested in these types of simulations. In particular, climate risk insurance can become more relevant by combining the analysis of the causes of these yield losses with the forecast of the yield itself.
This information would allow them to identify the least productive areas, whose crops are subject to strong climatic hazards, and therefore to support farmers with offers of insurance and protection against financial losses adapted to these risks linked to climate change.
[1] In the United States, a county, in English: a county is a form of local government, a territorial division smaller than a state but larger than a city or a municipality, in a state or a territory.
Predict and Decide