Improved supply intelligence for the global agriculture markets
Nick Kouchoukos and Corey Cherr, Thomson Reuters Lanworth
www.lanworth.com
FARMD (March 2012) | The characteristic volatility of agriculture markets is due largely to poor information, most often on the supply side. Turbulent weather and changing production practices in response to rising demand have complicated the already difficult task of predicting the future availability of crop resources at early stages in their production cycles. Over the past decade, Thomson Reuters Lanworth has applied the theories and methods of modern geographic and agronomic science to the ancient and persistent challenge of crop forecasting, with the goal of improving the efficiency of agriculture markets and supply chains worldwide.
Daily, global satellite imagery, improved long-range weather projections, and biologically realistic crop growth and yield models offer powerful new means to exceed the limits of conventional forecast methods. Foremost among such limits is sampling: the scale, expanse, and complexity of farming have restricted observations to a quasi-representative part of the whole. Further, it takes time to gather a significant sample—sometimes as long as it takes a crop to grow and be harvested. And field measurements, however essential, are inherently retrospective: they describe what has happened to a crop but offer little about what can or will. Satellite imagery, on the other hand, presents a total and immediate view of the agricultural landscape, allowing planted area to be surveyed in its entirety and monitored over the growing season. Growth and yield models account for the effects of weather to date and provide robust frameworks for evaluating the effects of probable forward weather scenarios.
Lanworth’s forecast process begins well ahead of planting with general models that project production based on data at hand and regularities observed over time. Models of local crop rotation practices and trends, for example, allow next year’s plantings to be projected from last year’s. Past relationships between climate anomalies such as ENSO and yield deviations from trend set first expectations for local production. Imperfect as they are, such models formalize assumptions, establish parameters, and quantify uncertainties—outlining pathways to various production outcomes and comparing their probabilities. Once planting begins and crops develop and form their yield, satellite imagery provides effective means to validate planting and yield models. Though current satellite technologies do not allow direct early season mapping of crop types or yield potential, they are capable of rejecting unrealistic model scenarios and tightening production expectations. Finally, Lanworth incorporates field observations to refine models and update production estimates. Fieldwork is expensive and time consuming, but planting and production models direct observations to where they are most needed and most effective.
The accompanying figure presents a graphical summary of Lanworth’s recent South American soybean production forecasts (orange points). Early models indicating substantial risk of lower production due to ENSO (La Niña) induced drought conditions quickly matured into confident assertions of lower than expected production potential. Official government agencies were slow to recognize losses (grey points), which drove a substantial correction in world soybean prices. This example highlights the agriculture markets’ need for more accurate and timely crop production forecasts and the developing ability of modern science and technology to deliver them.
Figure caption: South American soybean production forecasts. Grey bars represent the uncertainty of Lanworth’s estimate.