Back to Press Room

•           Could solve a very expensive problem
•           Current methods rely on what may be a faulty assumption

A team of California State University, Fresno viticulture researchers is developing a more accurate mathematical model for predicting wine-grape crop yields. The effort stems from significant errors in predicting statewide wine grape yields, says Robert Wample, director of Fresno State’s Viticulture and Enology Research Center (VERC) and leader of the project. “The inability to accurately predict grape crop yield is a multimillion dollar-a-year problem for the U.S. wine and grape industry, according to industry experts,” says Mr. Wample.

In 2005, for example, the official wine-grape yield prediction was off by approximately 30 percent. More recently, 30 percent to 50 percent errors in yield estimates have occurred, according to industry reports. “For the grower, it results in not being able to accurately and efficiently plan the harvesting process. For the winery, the problem is in insuring sufficient fermenting capacity, chemicals, storage space, including barrels and ultimately glass for bottling, and planning to market the resulting wines,” says Mr. Wample.

Conventional methods for estimating crop yield for grapes include counting clusters from selected areas of the vineyard and recording the number of berries and weight of each cluster. This data is then extrapolated using formulas based on number of vines per acre, number of acres and other information to obtain estimated yield per vineyard. There are variations of this basic method, but most rely on what might be a faulty assumption, says Mr. Wample. “When sampling data such as grape clusters, which grow over time, the use of competition models must be considered,” he says. “The usual assumption is that neighboring measurements tend to be alike. In fact, due to plant competition for nutrients and light, sizes of neighboring clusters may be negatively correlated.”

Researchers hope to develop a new mathematical model and associated statistical methods to improve yield estimates. “Our goal is to obtain yield estimates accurate to within 5 percent,” says Mr. Wample. “Anticipated results are improved vineyard and winery operations management. The yield model, once validated, could be used as a computational engine in a simulation model to forecast yield for various climatology and vineyard management scenarios.”

http://www.centralvalleybusinesstimes.com/stories/001/?ID=12857