GFS Forecast Model: CropProphet Yield Forecasts

CropProphet quantifies the impact of weather on U.S. crop yield and production. This post discusses a new quantification: direct calculation of the corn and soybean yield impacts from the 12Z GFS forecast model. The yield forecast implied by the 12Z GFS is calculated in 5-day increments as the data becomes available from NOAA. The first view is the 1-5 day weather forecast impact, then the 1-10 day, and then the 1-15 day impact. The content in CropProphet updates automatically as soon as the weather forecast files are acquired from NOAA and processed by our models.

The weather modeling centers around the world start running their global weather forecast models every day based on global observations from 00 UTC (i.e. 00 Zulu or 00Z), 06Z, 12Z, and 18Z. 00Z is 8 PM ET/7 PM CT in the United States during the U.S. crop season. The model data becomes available several hours later and this information is used to update the CropProphet weather model yield and production forecasts each day.

GFS Forecast Model: Requests for Yield

Several customers have commented to us, “I really like that you calculate the yield forecast implied by the weather forecast models. It would also be valuable if you could do the same for the 12Z GFS.” The 12Z GFS, also known as the 12Z Operational GFS run, is a single high-resolution realization of the GFS model that is run by NOAA before the lower-resolution 12Z ensemble run is executed. The data from the 12Z Operational GFS is released by NOAA around midday, and hence it’s sometimes called “the lunch run”.

The timing of the availability of this new weather forecast model data often moves corn and soybean futures as the market incorporates this new information. However, the market tends to react to very basic information about the forecast, with most market participants simply making a guess as to the implication and impact on likely yields.

The new CropProphet feature uses the advanced CropProphet county-level regression models to convert the 12Z GFS forecast model data into yield and production forecasts in real-time as the data is released to the NOAA servers. The normal process for ensemble data is shown below. CropProphet calculates the crop forecast based on the weather forecast of each ensemble member, and then the final CropProphet forecast is obtained from the median of the distribution (i.e. the most likely outcome). Note that the ensemble mean forecast is often not a realistic outcome (for example, its precipitation field is highly smoothed), and so we do not calculate the crop forecast from the ensemble mean.

Diagram of the ensemble weather model corn yield forecast process
Diagram of the ensemble weather model crop yield forecast process

The same process is used to calculate the implied yield and production from the 12Z GFS Operational run. In this case, however, the forecast is based on only a single model realization, as shown below.

Diagram of the 12Z GFS Opertional model corn yield forecast process
Diagram of the 12Z GFS Operational model crop yield forecast process

In summary, CropProphet now converts the 12Z GFS model forecast data into the impact on corn and soybean yield and production in real-time. An example full calculation from June 25, 2019 is shown below.

Map of the 12Z GFS Operational Yield Forecast
Example 12Z GFS Operational Corn Yield Forecast

We are beta testing this functionality with a few select customers now. If you’d like to try this new feature once it’s released to CropProphet, please request a trial now and reference the 12Z GFS forecast model data when we are in contact with you.