Curve Estimation-Based Forecasts of Wine-Grape Crop Size

• Activity Number: 10
• Type: Invited
• Date/Time: Sunday, August 3, 2003 : 2:00 PM to 3:50 PM
• Sponsor: ASA, San Francisco Chapter
• Author(s): Michael E. Tarter*+
• Companies: University of California, Berkeley

Abstract

It is asserted that curve estimation techniques can contribute both to productivity-forecast accuracy and data usage efficiency. Much of classic experimental design methodology can be traced to the Rothamsted Experimental Station. The simple geometry of California's rows of grapevines can provide a similar testing ground. Besides geometric simplicity, an additional characteristic can facilitate grapevine productivity studies. This is the subdivision of the problem into two subareas, sample design and berry, or grape, growth prediction. The number of grape clusters and the number of berries per cluster can, for all extents and purposes, be treated as variates whose values are constant throughout the growing season. Yet, because of complex patterns of growth on a given vine and from vine to vine, sophisticated sampling methodology is required. Similarly, how the volume of a given grape changes over time may not be described well by some simple single-valued function. Nevertheless, even though both the sampling and regression aspects of harvest forecasting are both, in themselves, complex, they can at least be considered individually.