Update - Forecasting corn yields for Kansas for 2018
Corn yields on target to USDA-NASS projections
A new and last crop yield forecast was released last week on October 11 by the USDA-NASS reporting a decrease on corn yield at the state level (-1 bu/acre, averaging 130 bu/acre) relative to the past yield forecast (September 1, 2018), including the final statewide crop acreage (+200,000 acres, averaging 5.2 million acres total). For the full report published by USDA-NASS see: http://usda.mannlib.cornell.edu/usda/current/CropProd/CropProd-10-11-2018.pdf
In a previous issue of the K-State Department of Agronomy eUpdate (Issue 712, September 28), a new yield forecast tool was discussed for corn in the state of Kansas (project sponsored by Kansas Corn). This tool is primarily based on real-time satellite data, historical yield data at county-level, and prediction of current geo-location of cornfields across the state. To obtain more information about the K-State “Yield Forecasting Tool” (YFT), visit the previous a previous eUpdate article related to this topic (Issue 653, September 29, 2017). The primary steps for this tool, presented in a simplified approach, are highlighted in Figure 1.
Figure 1. Theoretical framework portraying the main steps involved in the development of forecasting corn yields for the state of Kansas. Steps: 1- Data collection; 2- Building and validating yield forecasting models (YFM); 3- Building and validating land layer for corn 2017 (similar process was followed for 2018); and 4- validation of previous years. Infographic developed by Ignacio Ciampitti, Rai Schwalbert, and Luciana Nieto, K-State Research and Extension.
One of the main complexities is the lack of knowledge on the current cornfield locations across the state; therefore, a complex statistical technique was employed (random forest prediction) to predict corn geo-locations across the state for the current season (Figure 2).
Figure 2. Geolocation for cornfields predicted by the Random Forest classification model for 2018.
An updated corn yield forecast was obtained for the state of Kansas via utilization of satellite imagery from planting until beginning of October for this current growing season. Based on the satellite yield model developed by our team, the state-level yield prediction is 130 bu per acre (Figure 3), right on target to the yield prediction released by USDA-NASS (130 bu/acre).
Figure 3. Forecasting corn yields derived from satellite data for the state of Kansas.
A new step on the Yield Forecast Tool: County-yield prediction
Since our last eUpdate article, our team worked on developing a yield forecast at the county-level. Final validation of the county-yield data will be done once the county-yield information is released by USDA-NASS. Based on the yield forecast model, even when the model presents a degree of error close to 20 bu/acre, it is expected that yield trends should follow the expected projections. Overall, lower yields are projected for the eastern part of the state relative to the 2017 growing season, with yields improving as you move toward the western part of the state (Figure 4).
Figure 4. Predicted (based on satellite imagery) versus observed corn yields (panel A), average-county yield data from USDA-NASS for the 2017 growing season (panel B), and average-county yield data estimated from satellite imagery for the 2018 growing season (panel C) for the Kansas corn crop.
The Yield Forecasting Tool (YFT) predicted an average state-yield value of 130 bu/acre and lower yields for corn in eastern Kansas and higher yields for the western regions. At the state-level, the YFT was precise in predicting corn yields as related to the last forecast released last week by USDA-NASS (Oct 11, 2018).
Stay tuned for the further details coming out concerning the YFT as we continue to incorporate other components such as integration of weather data and other tools to improve identification of crop growth stages during the corn growing season.
Ignacio A. Ciampitti, Crop Production and Cropping Systems Specialist
Rai Schwalbert, KSUCROPS Production, Dr. Ciampitti’s Lab
Luciana Nieto, KSUCROPS Production, Dr. Ciampitti’s Lab