Lanworth Inc, an Illinois-based data analysis firm, won the attention
of the U.S. grain trade last week by issuing 2007 U.S. crop output estimates
that turned out to be highly accurate.
(article continues below useful links)
To help with its inaugural forecast, Lanworth relied heavily upon satellite
photos of crops taken throughout the summer in the United States -- as
opposed to the common method used by other crop forecasters of sending
scouts into fields and interviewing farmers.
Evolving satellite-mapping technologies and increasing availability of
such tools may change the way other, more seasoned crop experts predict
crop output across the globe in the future, Nick Kouchoukos, agricultural
analyst and director of information services at Lanworth, told Reuters.
"Our methodologies have considerable potential, which we think we've
demonstrated this year, to anticipate the ways that the USDA will move
toward its final estimate," he said.
Lanworth predicted U.S. corn production at 13.11 billion bushels, well
under the average trade estimate of 13.466 billion. The U.S. Agriculture
Department surprised the market with a 13.318 billion bushel estimate,
up just 10 million from its previous month forecast.
Lanworth was closer with its soybean production forecast of 2.58 billion
bushels, just under USDA's 2.598 billion figure and well below the average
trade estimate of 2.648 billion.
Lanworth expects the USDA to reduce its estimate of U.S. corn production
in its next report on Nov. 9, bringing the forecast closer to Lanworth's
projection.
"Satellite imagery has been used for a long time in agricultural
forecasting. What differentiates us from USDA and other forecasters is
that we are driven first and foremost by satellite imagery," Kouchoukos
said. USDA's massive network of crop experts allows it to nail down crop
yields extremely accurately, but the process takes months to complete.
Corn and soybeans are largely harvested by November, but USDA does not
release its final production figures until early January.
Companies like privately held Lanworth try to estimate production sooner
and with far fewer resources.
Itasca, Illinois-based Lanworth relies primarily on analysis of satellite
images of crops to determine planted acreage and the conditions under
which those crops grow. It supplements that data with analysis of soil-quality
maps and weather maps, as well as precisely targeted field sampling to
forecast crop yields.
SPOTTING VARIABILITY EARLY
A few key elements of Lanworth's forecasting method gave it an advantage
over other prognosticators that rely more heavily on in-the-field sampling
and farmer input.
"There was a combination of very optimistic information that had
been put out by USDA in September coupled with early yield reports in
September and early October that led the trade to expect that USDA would
be ever more optimistic in the October report. We didn't see it that way,"
Kouchoukos said.
Wet weather extended the planting window this spring and led to highly
variable yields in many key growing areas as crops were subject to different
growing conditions at various stages of development, he said.
"We can quantify variability more quickly and arrive at an estimate
of that variability and its effects more quickly than sampling-based schemas
can," Kouchoukos said.
Before merging with Westervelt Company and changing its name, Lanworth
was known as Forest One, a provider of land management services mainly
to the forestry sector. It has only recently expanded in other sectors,
including agriculture.
After two years of development, this is the first year that Lanworth has
forecast agricultural yields, starting with U.S. corn, soybeans and wheat.
It plans on expanding its coverage in the next year or two to include
wheat production estimates for other countries, as well as palm oil production
forecasts.
