Data for the forecasts are collected in two separate surveys conducted roughly in the last week of July and the first week of August for the August report.  

The Agricultural Yield Survey (AYS) queries farm operators in 32 states for corn and 29 states for soybeans asking operators to identify the number of acres to be harvested and to forecast the final average yield.

The sample of operators is based on a sophisticated sample design to achieve the desired sample size, and each state is expected to achieve a minimum response rate of 80 percent.

In 2011, approximately 27,000 operators were surveyed for all crops for the August report. Each operator is surveyed in subsequent months to obtain new forecasts of acreage and yield.

Historical relationships indicate that respondents tend to be conservative in early forecasts of final yields (to underestimate yield potential), particularly in drought years. This tendency is quantified and factored into official yield forecasts.

The second survey is the Objective Yield Survey (OYS) and is based on an area frame sample of cultivated land in 10 principal states for corn and 11 states for soybeans.  

Based on the June acreage survey, a random sample of fields is drawn in each state and enumerators visit these fields to take measurements needed to forecast yields in predetermined segments of the fields.

In 2011, these measurements were taken in 1,920 corn fields and 1,835 soybean fields.  

For corn, the number of plants and number of ears per segment are counted, and the size of the ears is measured.

For soybeans, enumerators measure row width and count the number of plants, number of main nodes, lateral branches, dried flowers and pods, and pods with beans in each segment.

The data are used to forecast grain weight (yield) per acre.

Each segment is visited in September, October, and November (if not yet harvested) to take new measurements and counts to form new yield forecasts.

Just before the operator harvests the field, each segment is hand harvested and weighed. The data from the two surveys are combined to forecast average yield and production.  

Good said that the USDA makes new yield and production forecasts in September, October, and November with final estimates released in January.

“Because yield forecasts are limited by crop maturity and are influenced by subsequent weather, the August yield forecasts tend to have the largest deviation from the final estimates released in January,” he said.

For the period 1970 through 2010, the August yield forecast for corn ranged from an over-estimate of 18.3 percent to an under-estimate of 10.1 percent.

The middle 50 percent of the forecast errors ranged from an over-estimate of 1.3 percent to an under-estimate of 5.1 percent.  

For soybeans, the August yield forecast ranged from an over-estimate of 16.2 percent to an under-estimate of 10.2 percent.

The middle 50 percent of the forecast errors ranged from an over-estimate of 3.0 percent to an under-estimate of 4.5 percent.