This summer, a Chicago, Ill.-based company predicted that U.S. corn crop production would eventually decline from USDA's Sept. 12 crop production estimate of 13.31 billion bushels. This prediction was based not on surveys, ear counts or a crystal ball, but on infrared satellite imagery of U.S. corn fields.
Six weeks later, USDA indeed lowered its corn production number to almost exactly what the company had predicted, thereby lending credence to crop imaging as a yield prediction tool.
Nick Kouchoukos of Lanworth, Inc., stressed that his company is not trying to do USDA's job — indeed, it relies on the government agency to provide a baseline for its predictions. Rather, its conclusions in 2007 simply predicted where USDA numbers might eventually end up.
“We are not by any means trying to develop numbers, or a whole new way of reporting that is different from USDA,” Kouchoukos said. “In fact, we begin by accepting that by the time of USDA's end of the year report, their method has given us something that is about as good as it's going to get.”
Since USDA's methodology is based on sampling, “it takes them a while to get to the right number,” Kouchoukos explained. “That's just because there is a human chain of information and you have to get to the right sample size before you can really constrain the error.
“This is where satellites have a real opportunity to make a difference. Every day, we get two images of the entire United States. That means we can assess crop conditions and changes from day to day very rapidly. We can plug them into our own models and can deliver a number that can anticipate the direction that USDA is going to move by several weeks.”
This year, satellite imagery pegged an accurate estimate for corn crop production six weeks earlier than USDA.
In September, USDA came out with an upward revision of its August numbers for corn production at 13.31 billion bushels. Shortly thereafter, Lanworth had arrived at what effectively were its final numbers for corn. In early October, Lanworth released an estimate for a corn crop of 13.14 billion bushels.
In its October and November reports, USDA started to revise its numbers downward, pegging corn production at 13.1 billion bushels in October and 13.17 billion bushels in November.
USDA soybean numbers were also revised downward and as of late November were spot on with Lanworth numbers.
Lanworth's methodology is not to focus on areas of the United States where crops are doing well, but on areas where definite problems exist. For example, for the corn and soybean crops, Lanworth used satellite imagery, soil and weather data, crop growth models and extensive field checks to project that localized drought conditions and variability in planting dates had limited the yield potential of the U.S. corn and soybean crops.
“If everything is on track, we just don't worry about it,” Kouchoukos said. “If it looks like it's not on track, there is a problem developing or if the crop looks a lot greener than it should be, we'll focus on that particular area and perhaps send a field team out to look at it. So we really rely on the satellite to tell us where everything is okay so we can focus our human energy just on the areas that really need (attention).”
Lanworth uses visible infrared imagery “which is directly related to chlorophyll content and other sensors that use the short-wave infrared, which can be very sensitive to soil conditions. We also use thermal infrared, which can tell us about heat exchange, and can tell us about temperature stress.”
Early in the season, Lanworth makes maps of crop type and condition on 200-meter grids across the United States. “Our routines allow us to do that pretty accurately. Then as we step through the season, we look at an image everyday to determine how a crop is developing.”
Kouchoukos says that the company is looking into “partnership opportunities” with USDA, “although right now, it's important for us to remain completely independent. We want our number to be a complement to theirs.”
Currently, the company is mapping out winter wheat fields that are beginning to emerge and next year, will move to cotton forecasting.
“That's not to say we will have a product spanning the entire 2008 season,” Kouchoukos said. “But we will be turning our development to cotton. We'll have some things to say about production next year, but it probably won't be until 2009 that we have a product of the quality that we offer for corn and soybeans.”
Lanworth employs 25 people with offices in Chicago, Ill., and Jackson, Miss. Last year, the company made its national figures on corn and soybean production available to the public and will likely do so again in some form for 2008.