Businesslink

Imperfect Predictions

Estimating yields is hardly a simple process, whichever method you choose. The third annual DTN/Progressive Farmer Digital Yield Tour, powered by Gro Intelligence, is a case in point.

On Aug. 10, the first day of this year's tour, a derecho devastated a wide swath of some of Iowa's most promising corn acres. Our tour simply couldn't account for this. Even though Gro's yield models, which informed our stories, updated daily, there wasn't enough time for the damage to register on one of its main inputs, the normalized differential vegetation index (NDVI).

P[] D[x] M[x] OOP[F] ADUNIT[] T[]

NDVI maps illustrate how green crops appear from space compared to a 10-year average, and it only updates once every two weeks or so. Damage takes time to affect plant health and a field's coloring.

In the Digital Yield Tour's first year, we saw a crescent-shaped brown patch on an NDVI map of Nebraska, which turned out to be hail damage from earlier that summer. It took a month for the hail scar to be visible from space.

Still, the progression of Gro's county estimates in Iowa is beginning to reflect the damage. Jasper County, hit by some of the strongest winds in the derecho, had the highest county average yield when we conducted the tour, 209.8 bushels per acre. Two weeks after the storm, Gro's models shaved off 9 bpa, and of this writing, yields have dropped to 198 bpa.

Gro's yield models become more accurate as the crop progresses. By mid-November, Gro's forecast for a national average yield has been within 3 bpa of USDA's final yield estimate on corn and 1 bpa on soybeans.

Yield monitors may have the final say, but crop models can provide the big picture much sooner, even if it's not immediate.

> Read Katie's business blog at about.dtnpf.com/business.

> You may email Katie at katie.dehlinger@dtn.com, or visit @KatieD_DTN on Twitter.

[PF_1020]