At the Kagoshima Prefecture Tea Market, the appearance and color of tea to be bid on are photographed with a digital camera for the purpose of improving the quality of crude tea, and the values obtained from texture and color analysis of the images are fed back to each farmer via smartphone, etc., along with the unit price and the image. In this study, we used various functions of JMP to examine a method to explain and predict the values of crude tea components, which have a significant effect on the taste and aroma of tea, from cultivation information and image analysis data.
We used a data set of 1,292 samples of first-season tea that were delivered to the tea market from each production area in the prefecture, image-analyzed, and then analyzed for components using near-infrared spectroscopy after being sold at auction. Principal component analysis using the component values of first-season tea rough tea (total nitrogen, free amino acids, theanine, fiber, tannin, caffeine, vitamin C) showed that 74.4% of the characteristics of the rough tea being auctioned could be explained by principal component 1, which is an index related to total nitrogen and fiber, and principal component 2, which is an index related to the amount of tannin and caffeine, and the unit price was positively correlated with total nitrogen and negatively correlated with fiber.
The control charts showed that total nitrogen and fiber often exceeded the control limits in the latter half of operation (mid- to late-season varieties). When the lower specification limit for total nitrogen was set at 5% and the upper specification limit for fiber was set at 22%, the non-conformance rates were 4.9% and 6.4%, respectively.
In order to predict total nitrogen and fiber from crude tea image analysis data, PLS regression and response surface model fitting were performed using cultivation information and 10 image analysis items as parameters. As a result, both component values could be explained by a model that incorporated the image analysis item "white stem" (an index of the maturity of the sprouts to be picked) in addition to the tendering date and variety. Furthermore, the design space function of the profile predicted that the total nitrogen and fiber specifications could be achieved at 98.9% by "picking and producing white stems of major mid- to late-ripening varieties to a level of 3 or less by May 7."
In conclusion, the values of the components in the crude tea can be predicted from the cultivation information and image analysis data, and the cultivation index for maintaining the values within the specifications can be obtained. We introduce a case where this information is used to provide guidance on the picking and processing of the tea.
