基于物候期气候因素的茶叶产量模型构建与种植管理

Construction of the tea yield model and planting management strategies based on climatic factors during phenological periods

  • 摘要: 本文聚焦于茶叶产量与符合茶树生长节律的物候期气候因素之间的复杂关系, 采用茶树物候期气候指标作为因子, 探索茶叶产量模型的构建, 并基于模型主要气候因子制定具有区域气候适应性的茶树种植管理策略。基于1990—2022年福鼎市气象观测资料和茶叶产量资料, 采用线性趋势法研究气候变化趋势, 运用多元线性回归法、主成分回归法及神经网络法3种方法构建福鼎市茶叶产量模型。研究结果表明: 1990—2022年福鼎市年平均温度以0.436 ℃∙(10a)−1的速率快速提升, 年日照时数与年降水量在不规律波动中分别以4.078 h∙(10a)−1和31.105 mm∙(10a)−1的速率提高, 气候变化影响茶叶产量的稳定性; 从不同茶树物候期气候指标形成的168个物候期气候要素中, 筛选出14个关键气候因子, 基于这些关键因子构建的神经网络产量模型的均方根误差(RMSE)为7.5956, 归一化均方根误差(NRMSE)为0.0807, 决定系数(R2)为0.9351, 平均拟合准确率(P)为93.13%, 该模型为最优产量模型。最后, 围绕最优模型中全物候期降水日数、春茶采摘期平均气压和第2次生长期≥10 ℃积温3个较高重要性水平因子, 新梢萌芽期蒸发量、休眠期蒸发量、休眠期平均风速、第2次生长期平均风速和第2次生长期平均日最低气温5个中等重要性水平因子, 结合当地气候变暖趋势, 进行分析探讨, 提出优化水分、土壤保墒、通风降湿、气象灾害防治和分散气候风险等契合福鼎市气候特征的茶树种植策略。

     

    Abstract: This study focuses on the complex relationship between tea yield and the phenological climatic factors that conform to the growth rhythm of tea trees. Using climatic indicators of the phenological periods of tea trees as factors, this study explores the establishment of a tea yield model and formulates tea planting and management strategies adapted to the regional climate based on the primary climatic factors identified by the model. Based on the meteorological observation data and tea yield data of Fuding City from 1990 to 2022, the linear trend method was adopted to study the climate change trends. Three methods, namely, multiple linear regression, principal component regression, and neural networks, were used to construct the tea yield model for Fuding City. The research results showed that from 1990 to 2022, the annual average temperature in Fuding City increased significantly by 0.436 ℃∙(10a)−1, and annual sunshine hours and precipitation increased by 4.078 h∙(10a)−1 and 31.105 mm∙(10a)−1, respectively, with irregular fluctuations. Climate change affected tea yield stability. Among the 168 phenological elements formed by the climatic indicators of different tea plant phenological periods, 14 climatic factors were identified. The evaluation results of the neural network yield model constructed based on these key factors showed that the root mean square error (RMSE) was 7.5956, the normalized RMSE (NRMSE) was 0.0807, the coefficient of determination (R2) was 0.9351, and the average fitting accuracy rate (P) was 93.13%, making it the optimal yield model. Finally, focusing on the three high-importance factors in the optimal model, including the number of precipitation days during the full phenological period, the average air pressure during the spring tea picking season, and the accumulated temperature ≥ 10 °C during the second growth period, as well as the five moderate-importance factors, including evaporation during the bud germination period, evaporation during the dormancy period, average wind speed during the dormancy period, average wind speed during the second growth period, and average daily minimum temperature during the second growth period, combined with the local climate warming trend, an analysis and discussion were conducted. Tea planting strategies adapted to the climatic characteristics of Fuding City have been proposed, such as optimizing water resources, conserving soil moisture, using ventilation to reduce humidity, preventing meteorological disasters, and dispersal of climate risks.

     

/

返回文章
返回