A multi-dimensional investigation into white tea produced in northern China has revealed how three years of controlled storage systematically reshape the beverage’s sensory profile, optical properties, and chemical matrix. Researchers analyzing population-cultivar white tea from Junan County in Shandong Province tracked the molecular trajectories of leaves stored for zero, one, two, and three years under dark, temperature-controlled conditions. The findings demonstrate that prolonged storage triggers a substantial decline in bitter and astringent catechins, notably reducing total tea polyphenols from roughly seventeen percent of dry leaf weight in newly manufactured samples down to approximately ten point six percent after three years. Simultaneously, free gallic acid accumulated across the storage period as ester-type catechins underwent hydrolytic cleavage, softening the beverage’s initial astringency into a mellower, fuller sensory expression. By integrating biomimetic taste sensing, colorimetric measurements, liquid chromatography, and machine learning, the researchers constructed a detailed baseline of storage-induced aging in high-latitude tea, while demonstrating both the utility and boundaries of computational modeling.
White tea has traditionally been cultivated in southern Chinese provinces such as Fujian, where manufacturing relies on minimal processing consisting purely of natural withering and drying without rolling or pan-firing. In recent decades, agricultural programs have introduced tea cultivation into northern latitudes such as Junan County, where large diurnal temperature swings and extended frost-free periods yield fleshy tea leaves naturally rich in polyphenols and free amino acids. To assess how this northern material ages over time, researchers collected fresh tea leaves consisting of one bud and two leaves from a single experimental plot across four consecutive spring harvests from 2022 to 2025. Each annual cohort underwent an identical processing protocol at the same manufacturing facility before being sealed in light-impermeable aluminum foil laminate bags. The packaged lots were stored under constant conditions of twenty-five degrees Celsius and fifty-five percent relative humidity within the Shandong Academy of Agricultural Sciences, establishing twelve independent experimental units across four storage intervals.
Spectrophotometric and colorimetric evaluations showed that tea infusions underwent profound macroscopic optical transformations across the three-year timeline. Newly processed tea produced liquor characterized by higher haze and turbidity values of five point thirty-seven and fourteen point ninety, respectively, whereas infusions prepared from three-year-old tea exhibited significantly reduced haze of one point eighty-eight and turbidity of nine point twenty-seven. This reduction in suspended particulate haze imparted greater macroscopic visual clarity to the aged liquor. Conversely, the overall light transmittance of the liquor experienced a slight but statistically significant drop, decreasing from eighty-six point eighty-three percent in fresh samples to eighty-three point forty-two percent in three-year-old samples. The team attributed this optical divergence to the formation of high-molecular-weight colored pigments, including theaflavins, thearubigins, and theabrownins generated through non-enzymatic catechin autoxidation, which absorb visible light in the blue-violet spectrum and deepen the liquid’s visual tone.
Objective colorimeter parameters captured the progressive darkening and chromatic migration of the brewed tea over time. Instrumental lightness decreased in an approximately linear manner, dropping from ninety-four point sixty-six in unaged tea to ninety-one point fifty-three in tea aged for three years, confirming a darker visual character. Measurements along the red-green coordinate revealed a slight enhancement in greenness during the initial year of storage, shifting from minus two point fifty to minus three point twenty-one before stabilizing. Concurrently, the yellow-blue coordinate shifted dramatically in the positive yellow direction during that first year, rising from nine point twenty-five to twelve point fifty-one. Saturation followed an identical early trajectory, surging from nine point fifty-nine in new tea to twelve point ninety-two in one-year samples before plateauing. Total color difference increased significantly during the first year, establishing that the primary chromatic reorganization of the liquor occurs early in the storage lifecycle.
Sensory profiling by a qualified panel of three trained tea tasters confirmed that storage transformed the liquor from sharp and astringent into rounded and mellow. Panelists scored five core attributes comprising appearance, liquor color, aroma, flavor, and infused leaf on a one-hundred-point scale. Newly manufactured tea achieved scores of ninety or above across all categories, excelling in leaf freshness and tenderness with an infused-leaf score of ninety-two. As aging progressed, appearance and infused-leaf scores gently drifted downward to ninety, while flavor ratings climbed from ninety to ninety-one. Liquor color ratings peaked in the second year of storage at ninety-four, lifting the composite sensory score from ninety-one point fifteen in new tea to a peak of ninety-two point twenty at year two, followed by a slight decline to ninety point sixty-five at year three. The aroma evolved sequentially from honey and pekoe notes in new tea to fresh pekoe at year one, floral and fruity accents at year two, and distinct medicinal notes by year three.
Biomimetic quantification via an electronic tongue equipped with lipid-membrane sensors provided objective electrical measurements that corroborated the human tasting scores. Unaged tea registered the highest intensities for bitterness, astringency, and sourness, as well as significantly higher aftertaste responses for astringency and bitterness than any aged sample. Over the storage window, measured bitterness declined precipitously to zero point forty-two, while astringency decreased to eight point sixty-three by the third year. Umami intensity, which stood at twenty-three point zero seven in unaged tea due to its high initial pool of free amino acids, diminished steadily to fourteen point zero six at year three. Crucially, this drop in umami did not leave the brewed tea tasting thin or insipid; the sensor measuring physical richness and body increased from zero point thirty-six in fresh tea to between zero point ninety-five and one point twenty-seven across the aged samples, chemically explaining the sensory impression of enhanced roundness.
Chromatographic analyses resolved the individual chemical compounds responsible for these profound changes in taste and appearance. Total tea polyphenols declined significantly over three years, dominated by the degradation of catechin monomers. Among individual flavanols, the major ester-type catechin epigallocatechin gallate plummeted from approximately eight point zero percent of dry weight in unaged leaves to around three point zero percent after three years. Epimerization simultaneously converted a portion of epigallocatechin gallate into gallocatechin gallate, which rose significantly in third-year samples. In mirror image to the degradation of esterified catechins, free gallic acid remained stable during the initial twelve months before climbing sharply to one point ninety-seven percent at year three as ester bonds underwent continuous hydrolysis. Flavonoids decreased from seven point eleven percent to three point eighty-four percent, and soluble sugars decreased moderately, while caffeine concentrations displayed no statistically significant change over the three-year window.
To understand the speed of these chemical transformations, researchers applied apparent first-order kinetic modeling to constituents exhibiting monotonic downward trends across the sampling points. Goodness-of-fit assessments identified six compounds exhibiting coefficients of determination of zero point eighty or higher, making them suitable for parameter estimation. Degradation rates varied substantially across specific molecular structures rather than conforming strictly to whether a catechin was esterified. Catechin C degraded the most rapidly, exhibiting an apparent reaction rate constant of zero point seven hundred ten per year and an estimated half-life of roughly zero point ninety-eight years. Among esterified catechins, epicatechin gallate degraded with a rate constant of zero point three hundred sixty-one per year, while epigallocatechin gallate decayed at zero point three hundred thirty-five per year. Epigallocatechin decayed most slowly at zero point two hundred forty-eight per year, indicating that the presence of an ortho-trihydroxyl group on the B ring and galloyl groups governs oxidative susceptibility.
Statistical evaluation of competing kinetic frameworks revealed critical methodological boundaries in determining the true mathematical order of white tea degradation. Comparing zero-order, first-order, and second-order kinetic models using small-sample corrected Akaike information criteria established that four temporal sampling points are statistically insufficient to discriminate among reaction orders. For all nine decreasing chemical constituents, differences in corrected Akaike scores between the top two models remained below the standard evidentiary threshold of ten, with zero-order equations showing nominally lower scores for seven compounds. Even when individual batch observations were tested separately, only epigallocatechin gallate and epicatechin gallate produced distinguishable model separations, both marginally favoring zero-order kinetics. Consequently, the researchers emphasized that the calculated first-order rate constants and half-lives represent descriptive, apparent kinetic estimates under specific environmental conditions rather than a definitive determination of true chemical reaction order.
Supervised multivariate statistical modeling using orthogonal partial least squares discriminant analysis confirmed clear separation among all four storage cohorts. A global multi-class model accounted for ninety-nine point seven percent of the variance in explanatory variables and ninety-nine point one percent of class variance, supported by cross-validation and rigorous two-hundred-iteration permutation tests that yielded negative quality intercepts. Pairwise comparison models pinpointed which variables drove class distinction at different stages of maturity. In the comparison between fresh tea and one-year-old tea, platinum-cobalt color units served as the sole verified differential marker meeting both variable importance thresholds and false-discovery-corrected significance, confirming that optical darkening dominates early aging. In contrast, distinctions between second- and third-year teas were driven primarily by taste attributes, notably declines in electronic-tongue saltiness, sourness, umami, and concentrations of flavonoids and epigallocatechin gallate.
Network analyses combining Mantel tests and Pearson correlations elucidated the structural interdependencies connecting chemistry, optical parameters, and taste. Total polyphenols and gallic acid functioned as opposing organizational hubs across the correlation matrix, reflecting the oxidative depletion of flavanols against the hydrolytic accumulation of ester cleavage products. Total polyphenols correlated positively with soluble sugars, principal catechins, liquor lightness, and sensory astringency, demonstrating how polyphenol loss softens brew intensity. However, when researchers calculated partial correlations while statistically controlling for storage duration, the vast majority of these strong pairwise associations collapsed toward zero. For example, the correlation between bitterness and epigallocatechin gallate plummeted from zero point nine hundred fifty-three to zero point zero twenty-five. This statistical collapse demonstrates that the apparent coupling among color, taste, and chemical concentrations represents concurrent, parallel consequences of storage time rather than direct mechanistic dependencies between individual indicators.
To evaluate predictive tools for extended storage, the researchers implemented a physics-informed neural network incorporating four hidden layers and soft first-order kinetic constraints. In cross-validation tests interpolating within the zero-to-three-year window, the physics-informed model exhibited no accuracy advantage over conventional first-order equations or unconstrained data-driven neural networks, yielding root-mean-square errors comparable across all architectures. However, in sparse extrapolation challenges trained on extreme single-year subsets, purely data-driven networks suffered catastrophic instability, producing errors roughly fifteen times greater and generating unphysical negative concentration values. The physics-informed network maintained smooth, plausible decay curves. Nonetheless, the authors cautioned that computational extrapolations spanning five to ten years of aging remain strictly illustrative simulations lacking external validation against physical tea samples, and must not be used as commercial verification standards until older, verified northern harvests become available for experimental analysis.
The authors explicitly outlined several scientific limitations inherent to their study design that constrain how broadly these findings can be generalized. Because the experimental design relied on three independent batches harvested across four successive years, storage duration remained completely collinear with agricultural production year, leaving open the potential influence of annual climatic variations during leaf growth. Furthermore, sample sizes provided limited statistical power for equivalence testing, meaning that non-significant changes in caffeine and free amino acids cannot be formally interpreted as proven biochemical stability. Future investigations expanding sample sizes across controlled temperature and humidity grids will be essential to establish fully validated kinetic standards for the expanding northern white tea industry.
Subject of Research: Chemical, optical, sensory, and kinetic transformations of Shandong white tea during storage
Article Title: A multi-dimensional analysis of quality changes in Shandong Junan white tea during storage
Article References: Wang, D., Dong, C., Zhang, F., Guo, M., Chen, Z., & Zhang, X. (2026). A multi-dimensional analysis of quality changes in Shandong Junan white tea during storage. Food Chemistry: X, 40, Article 104770. https://doi.org/10.1016/j.fochx.2026.104770
Image Credits: AI Generated
DOI: 10.1016/j.fochx.2026.104770
Keywords: White tea, Junan County, Shandong agriculture, Tea storage, Polyphenol degradation, Catechin kinetics, Epigallocatechin gallate, Gallic acid, Electronic tongue, Spectrophotometry, Sensory evaluation, Food chemistry, OPLS-DA, Physics-informed neural networks, Reaction kinetics, Tea quality grading, Flavonoids, Soluble sugars
News Source: Bethany Barker. (November 12, 2026). Storage Transforms Shandong White Tea Flavour While Preserving Catechin Oxidation Pathways. Scienmag.
