Accurately simulating occurrence dates of spring flowering and autumn leaf coloration of ornamental plants is of significant importance for revealing phenological response of urban vegetation to climate change and predicting the optimum timing for flower and foliage viewing. This study employed the Unified Forcing Model (UniForc) and Unified Chilling Model (UniChill) to fit the first flowering and 50% flowering dates of 20 plant species from 1979 to 2019 in Beijing Botanical Garden, and the Low Temperature and Photoperiod Multiplicative Model (TPM) to fit the first leaf coloration and 100% coloration dates of 10 plant species. The errors of optimum models in simulation and prediction were evaluated. Results show that spring flowering of ornamental plants is mainly driven by forcing temperature during ecodormancy and growth periods, but less restricted by chilling temperature during endodormancy period. First leaf coloration of ornamental plants is mainly driven by the process of leaf senescence induced by daily minimum temperature decrease, and 100% leaf coloration is mainly driven by the process of leaf senescence induced by photoperiod shortening. The average simulated root-mean square errors (RMSE) for first flowering and 50% flowering are 3.7 days and 3.2 days, respectively, and simulated RMSEs for the two spring phenophases show good interspecific synchronization. The average simulated RMSEs for first leaf coloration and 100% leaf coloration are 9.4 days and 5.6 days, respectively, but simulated RMSEs for the two autumn phenophases do not display interspecific synchronization. Simulated RMSEs of flowering dates and leaf coloration dates of various plants correlate significantly and positively with their standard deviation of interannual variations. The simulated and extrapolating RMSEs of optimum spring and autumn process-based models are very close, indicating that the models have high robustness. 

timing sequential beauty of urban landscape, first and 50% flowering, first and 100% leaf coloration; climate driving, process-based simulation and prediction

为了准确地模拟观赏植物春花期和秋叶期的发生时间, 揭示城市植物物候对气候变化的响应特征并预测赏花和观叶的最佳时机, 利用统一促进模型、统一冷激模型与日最低温–光周期耦合模型, 分别模拟北京植物园1979—2019年20种春花植物开花期和10种秋叶植物叶变色期的年际变化, 并评价最优模型模拟与预测的误差。结果表明, 园林观赏植物春季开花期的发生主要受生态休眠期和生长期促进温度驱动, 内休眠期冷激温度需求没有对春季开花期的发生起到限制作用; 观赏植物秋季叶始变色期的发生主要受温度降低诱发叶片衰老的过程驱动, 叶全变色期的发生主要受光周期缩短诱发叶片衰老的过程驱动。开花始期和开花盛期模拟的平均均方根误差(RMSE)分别为3.7和3.2 d, 且这两个物候期的模拟误差具有较好的种间同步性; 叶始变色期和叶全变色期模拟的平均RMSE 分别为9.4和5.6 d, 而这两个物候期之间的模拟误差不具有种间同步性; 物候模型对各种植物开花日期和叶变色日期模拟的RMSE与其年际波动标准差呈显著的正相关关系。最优春季和秋季物候过程模型的外推检验与模拟检验的RMSE数值相近, 表明模型具有较高的稳健性。