With Soybean Yields a Mystery to Many, Here’s What Gro’s Model Predicts

04 September 2019

US crops face greater than usual uncertainty this year, and none more so than soybeans.

The USDA’s WASDE report for September, due out next week, will be the department’s first reliable indicator of yield for the nation’s soybean crop. The estimate will be based on the department’s first objective yield survey of the season, which gathers on-the-ground measures of crop progress and is considered the most trustworthy data.

In August, the USDA NASS reported that the area planted to soybeans was down 14% from last year, as farmers shifted a surprisingly large number of acres to planting corn. But in the absence of new data, the USDA last month left unchanged its soybean yield estimate, which was last updated in June. Yields will certainly decline this year, but by how much is the key question.

Ahead of next week’s WASDE report, this Featured Insight checks in with Gro’s US Soybean Yield Model, which is currently forecasting yield declines from 2018 in 76% of soybean producing counties. Using a number of data sources on the Gro data platform, we analyze growing conditions down to the county level to explain why yield forecasts are varying so widely from state to state and what this will mean for currently record high US soybean stocks.

Read more:

The Season So Far
Struggling Counties
Optimistic Areas
Outlook

sample image desc
The choropleth map on the left (green) shows the Gro US Soybean Yield Model final estimates for the 2018 crop at the county level. The map on the right (red) shows the current Gro yield estimate for the 2019 crop at the county level. Areas with darker shading indicate counties where yields are expected to be high, and areas with lighter shading indicate counties where yields are expected to be low. When comparing the two maps, it’s easy to notice in the map on the right the lighter shaded areas that were hit harder with prevent plant and late planting, like in northwest Ohio. Click on the image to go to an interactive map on the Gro web app.

The Season So Far

Drenching rains and flooding got planting off to a late start across much of the Midwest. That should have boosted acreage of soybeans, which go in the ground later than corn. Instead, soybean acreage plunged 14% to 76.7 million acres, as farmers reacted to a late corn price spike to plant more corn. In addition, a record number of prevent plant acres in several states reduced acreage from both crops.

Soybeans have looked disappointing for many months, as the China/US trade war, now in its second year, slashed exports and depressed prices. As a result, US soybean stocks have soared 144% since last year to 29.1 million tonnes, the highest level ever.

sample image desc
Record high soybean stocks accumulated as a result of the China/US trade war. But reduced soybean acreage and a poor yield forecast in 2019 may cause soybean stocks to drop back to more normal levels in just a single season. Click on the image to go to an interactive display on the Gro web app.

Late planting means many soybean plants are behind in reaching maturity, confounding efforts to estimate yield. In June, the USDA estimated average national yield would fall from 51.6 bushels per acre in 2018, the second highest on record, to 48.5 bushels per acre this year. The August WASDE report reiterated the June yield estimate. But that figure is widely expected to be reduced in the September WASDE report next week.

Gro’s US Soybean Yield Model predicts a less optimistic harvest. The model, which updates daily, estimates total US soybean yield of 46.9 bushels per acre as of September 4. That, combined with reduced soybean acreage, implies total soybean production of 3.6 billion bushels, a steep 21.7% decline from last year. It also suggests that bloated US soybean stocks could be drawn down sharply this year by 437.5 million bushels, or 41%.

sample image desc
Despite rising estimates in recent weeks, Gro’s US Soybean Yield Model still forecasts a substantial decline in yields in 2019 (blue line) from a year earlier (green line). Higher NDVI measurements are the biggest contributor to recent gains in the yield outlook. Click on the image to go to an interactive display on the Gro web app.

Gro’s machine-learning-based yield model continuously analyzes 60 features, including satellite and climate data and plant-condition ratings. Digging into these features, we can gain insight into which variables are most affecting yield forecasts nationally and at the state and county levels.

One of the Gro model’s main variables pulling down yields for both soybeans and corn this year is poor vegetative health. NDVI, a measurement of vegetative health collected using satellite imagery, has been recorded at its lowest levels since 2012 for Illinois, Ohio, and Indiana.

sample image desc
Exceptionally wet soils in Illinois and throughout the Midwest impeded and prevented planting of both corn and soybeans this spring. Note the high readings in the chart for the March through May period. Late planting and planting in wet soils are partially to blame for the decrease in yields between 2018 and 2019. Click on the image to go to an interactive display on the Gro web app.

Crop condition ratings, released weekly by the USDA and inputted to Gro’s yield model, also have weighed on Gro’s soybean yield forecasts. In early August, soybeans rated in poor condition reached 18%, compared with 3% at the same time in 2018.

Exceptionally wet soil early in the planting season has also diminished soybean yield estimates. Beginning in mid-March and going through early June, soil moisture measurements were consistently at the highest levels when compared to the previous five years. Besides preventing planting, excessively wet soil is more readily compacted, impeding root development and depressing yields as the crop reaches maturity.

Struggling Counties

Some counties have been hit harder this year than others, especially in Illinois. Twenty-four counties nationally are expected to have yield declines upward of 20 bushels per acre; of those 24 counties, eight are in Ohio and 13 in Illinois.

Ohio was also hit hard with prevent plant acres and set back by late planting, especially in the northwest part of the state where farmers felt the brunt of the heavy rains. In the Gro model, the key factor pressuring yields is NDVI, with index measurements the lowest recorded in the past 20 years. Williams County, bordered by Indiana to the west and Michigan to the north, is predicted to be Ohio’s hardest hit county in 2019, with average yield of 36.1 bushels per acre, down 39.1% from last year. Overall, Ohio’s soybean yield is currently forecast to drop by 22% to 45.5 bushels per acre this year, according to the Gro yield model.

sample image desc
This chart shows the yield history of select counties in Illinois and Ohio as colored lines, while the bars show the current Gro SoybeanYield Model forecast for those counties. These are some of the biggest year-over-year declines in yield in the country. Click on the image to go to an interactive display on the Gro web app.

Illinois, the country’s biggest soybean-producing state for the past three years, will see average yield drop to 50.7 bushels per acre, or 16%, from a record high of 65 bushels per acre in 2018, the Gro model shows.

The hit to yields in Illinois is concentrated in the central part of the state. Sangamon County was the highest yielding county in all of the US in 2018, with a whopping 82.3 bushels per acre. In 2019, Sangamon is expected to yield 57.85 bushels per acre, according to the Gro model, a 29.7% decline from last year and the lowest figure since the 2012 drought.

By comparison, corn’s average national yield for this year is expected to be down 4.5% to 168.5 bushels per acre. But in states such as Ohio, Illinois, and Indiana, corn yields could fall as much as 24%, according to the Gro corn yield model. The USDA surprised markets on Aug. 12 when its WASDE report estimated corn planted area at 90 million acres and average corn yield of 169.5 bushels per acre. Both estimates were above market expectations and sparked a plunge in corn prices.

Optimistic Areas

It’s not all bad news across the US Midwest, however. In the western part of the Corn Belt, states including Nebraska, Missouri, and Iowa haven’t taken as big a hit to yield as have the eastern states and South Dakota. Soybean yield in Nebraska, for instance, is forecast to be just 3.5% lower than the 2018 number. This would make Nebraska the highest yielding state for soybeans this year. Some inputs to the Gro yield model, especially soil properties, have worked in favor of sandier states such as Nebraska and Wisconsin. Sandy soils drain water more quickly, which helps during an uncharacteristically wet year.

sample image desc
Counter to the overall trend, a few counties, including the ones in central Missouri included in this chart, are expected to have higher soybean yields this year, mainly because they had fewer setbacks during planting season. The line with markers shows favorable 2019 NDVI measurements for the four Missouri counties. Other lines represent NDVI from previous years. Click on the image to go to an interactive display on the Gro web app.

Some Midwest counties should expect to see higher soybean yields this year, the Gro model shows. Several counties in south-central Missouri, including Dallas, Morgan, and Laclede counties, could improve by more than 8 bushels per acre in 2019. These areas missed the brunt of the rainfall issues farther north, and look to be on track for a decent harvest.

Outlook

Soybeans still have a few weeks before harvest ramps up in the major producing states. NDVI has risen over the month of August to more normal levels, which may suggest some crop recovery before harvest. And both central Illinois and northwest Ohio, two of the weakest yielding areas, are forecast to receive some rains in the next week, which should assist late-season pod filling for the immature soybean crop. Gro Intelligence representatives who participated in this year’s Pro Farmer Midwest Crop Tour from Aug. 19-22 observed many fields of immature soybean plants that were still flowering, far behind the typical schedule.

As markets eagerly await USDA yield forecasts, Gro’s yield model is currently predicting an average national yield decline of 9% to 46.90 bushels per acre, compared with last year. On a state level, however, yield forecasts vary widely, ranging from a high of 57.33 in Nebraska to a low of 40.88 in South Dakota. Final estimates from the Pro Farmer Crop Tour include 39 bushels per acre in both Ohio and South Dakota, 50 bushels per acre in Illinois, and 46.1 bushels per acre nationally. Some counties in Ohio and Illinois are poised to see yields drop by half from last year. That, combined with a steep drop in soybean planted acreage, could draw down US stocks from their current historic highs to more normal levels in a single year.

Contact sales