Sorghumyield® App, a New Tool for Predicting Sorghum Yields: Validation and Calibration.

Sorghum yield estimation before harvest can be a tedious and erratic task, but sorghum key-stakeholders would like to know about the yield potential before harvest. Sorghum yield can be dissected into three major components: heads per acre, seeds per head, and final seed weight. The most variable yield component is the grain number per head. Thus, a close yield estimation on this component can assist producers in guiding the yield prediction process. Following this rationale, a project started 2-yrs ago individually screening >3,000 sorghum heads with imagery collection and grain counts. A multiple step approach was developed with the final objective of estimating not only final grain number (based on head size), but also predicting sorghum yields before harvest. An electronic application was developed (SorghumYield^®, Android) with the goal of facilitating the yield prediction process in farmer fields across the globe. The users provide field data such as row spacing (for counting sorghum heads), head number per unit of row-length. After this step is done, a mandatory CropYield sheet is needed so the users can proceed and take sorghum head images using the electronic device (e.g. smartphone). A grain number estimation will be gathered from each of the images processed by the App, a logarithm was created via head volume estimation and final grain number. The last steps are related to the grain weight and final yield estimation, which can be based on seven-day weather forecast plus soil information for this specific farm. During the current growing season, on-farm measurements were gathered to validate and calibrate the SorghumYield^® App. This step is essential for the technology transfer objective. The on-farm evaluation is planning to provide a response about precise timing on making yield predictions and accuracy of the SorghumYield^® App as compared with farmer combine yields at harvest.