By Jochum Wiersma
The USDA-NASS' July 1 yield forecasts for barley, oat, and spring wheat were 67, 71, and 61 bushels per acre, respectively. This would mean a new state record for spring wheat, while the forecast for barley and oats are 10 and 7 bushels off the records set in 2015.
To estimate yield the USDA-NASS collects farmer's assessment of yield prospects throughout the growing season, i.e. the USDA-NASS asks producers to predict their final yield. At first glance, this may seem a bit unscientific and not very accurate. The statistical methods that are used to crunch the collected data and have it confess a forecast, however, are robust and because enough producers are surveyed, the forecasts have been proven predictive at the aggregate level. This is, in a way, a testament that you each know you crop and operation pretty well. The completely methodology can be found here.
Estimating final yield in May and June in an individual field is fraught with a much larger error than at the aggregate level of a whole state. Once the crop, however, has headed and the grainfill has started it get much easier to estimate the final yield. Rather than just predicting the final grain yield, you can actually quantify the number of stems per unit area and the number of kernels per spikelet or panicle. This means that when you multiply this with an average kernel weight you can have a more accurate estimate of grain yield. The formulas for wheat, barley, and oats are:
Wheat: Grain yield (bu/acre) = (kernels per spike x spikes per 3 ft of row) x 0.0319
Barley: Grain yield (bu/acre) = (kernels per spike x spikes per 3 ft of row) x 0.0389
Oats: Grain yield (bu/acre) = (kernels per spike x spikes per 3 ft of row) x 0.0504
The multiplication factor for each crop is a function of bushel weight, average kernel weight, and row spacing. To adjust the yield estimate for other row spacing, simply divide the estimated yield by your row spacing (in inches) and multiply the product by 7.
A word of caution is that you need to have sampled a representative length of row and that the average kernels weight used in these formulas are the same as the eventual kernel weights. The first source of error can be addressed by taking multiple samples in a single field; the second source of variability is a bit more challenging to deal with. As calculating seeding rates have taught you over the years, the number of seeds per lbs. can vary considerably, not just between varieties but also from year to year.