中国北方温带秋季树木物候过程模型的构建与预测研究

Using first leaf coloration and leaf fall end data of six to eight tree species and daily temperature and wind speed data during 1981-2011 at 136 stations in the temperate zone of northern China, we will fit two existing process-based leaf senescence models and establish three new process-based leaf senescence models, and select optimum models for each species at each station. Then, we will examine performances of each optimum local species-specific model in predicting first leaf coloration dates at all external stations within the corresponding climate region of the study area and select the local species-specific models with maximum effective predictions as the regional unified species-specific models in each climate region. Further, we will substitute interpolated daily temperature data at 8 km×8 km grids into the regional unified species-specific models, and reconstruct spatiotemporal patterns of first leaf coloration dates of the selected tree species across northern China during 1960 to 2011. The same modeling process will be carried out at 8 km×8 km grids using the daily temperature data acquired from outputs of Regional Climate Models (RCMs) under the project Coordinated Regional Downscaling Experiment (CORDEX)-EAST ASIA. The model structure and parameters fitted by the two types of temperature data sets will be compared at station-grid levels in order to evaluate reliability of the scenario temperature data modeling. After that, we will reconstruct spatiotemporal patterns of first leaf coloration dates over 1950 to 2005 and predict spatiotemporal patterns of first leaf coloration dates over 2006 to 2100 under different climate change scenarios across northern China. In addition, we will also gather the first hand data through an in situ phenology and temperature observation along a vertical transect in the western mountain of Beijing to fit the above five process-based leaf senescence models. The in situ data based optimum models will also be compared with optimum models based on station phenology and temperature data in the Beijing area in order to examine similarities and differences between short-term spatiotemporal series of leaf coloration acquired from a vertical transect and long-term time series of leaf coloration acquired from a standard meteorological and phenological station in response to temperature.

利用中国北方温带地区1981-2011年136个站点6-8种树木的秋季物候数据和逐日气温与风速数据，拟合5种叶片衰老过程模型，并对模型进行优选。通过检验最优模型预测相应气候区内非建模站点秋季物候的表现，将具有最大预测能力的最优模型确定为区域统一模型。进而，将8 km×8 km格点气温插值数据输入区域统一模型，重建中国北方1960-2011年秋季物候的时空格局。同时，利用区域气候模式输出的8 km×8 km格点逐日气温数据进行上述叶片衰老过程的模拟，重建并预测中国北方1950-2005年和2006-2100年秋季物候的时空格局，并比较基于两种气温数据集的模型结构与参数，以评价情景气温数据建模的可靠性。此外，通过沿北京西山地区垂直断面的实地物候与气象观测收集第一手数据，拟合上述叶片衰老过程模型，验证取自短期观测的秋季物候时空序列与取自长期观测的秋季物候时间序列对气温响应的异同。

秋季物候仍是气候变化研究中相对薄弱的环节，这不利于全球碳循环及其对气候变化敏感性的精确评估。在收集107个站点、11种木本植物和17种草本植物的221个时间序列秋季物候数据的基础上，我们主要进行了树木叶变色日期与气象因子关系的统计分析和过程模型模拟。重要研究结果包括：（1）温度对叶变色的影响比干旱更为显著，并且生长季均温比秋季日最低气温对叶变色发生早晚的作用更为重要。较高的生长季均温和较低的秋季日最低气温会导致叶变色期的提前。此外，生长季均温还可以抵消触发叶变色的秋季最低气温的需求。我们的发现深化了对冬季落叶树木叶变色机理的认识，并指出叶片寿命控制（依赖于生长季均温）和秋季低温控制以及两者之间的相互作用是影响叶变色的主要环境诱因。在气候增暖的情景下，叶变色日期是否提前或者推迟主要取决于生长季均温对秋季低温抵消效应的强度。（2）与现有的秋季物候模型相比，增加生长季温度与干旱事件次数对叶变色日期影响的改进秋季物候过程模型的模拟效果最好。这表明气候变暖导致的叶变色期推迟可能会以复杂的方式受到生长季气候对叶变色时间影响的调节：较高的生长季温度会加速叶变色期的出现而生长季较少的降水则会推迟叶变色期的出现。（3）提出了一种基于光周期和低温耦合预测叶片衰老时间的全新过程模型，大大简化了前人的光周期和低温耦合模型。新模型将光周期和低温作为叶片衰老的独立条件，当二者之一达到阈值时，叶衰老过程即可启动，而叶片的衰老速率则由光周期与日最低气温乘积的指数函数予以描述。该模型在反映叶衰老生理生态机理的真实性、模拟与预测的准确性和模型结构的普适性方面均优于现有的模型。模拟结果表明，在夏季光周期较长的北方，叶衰老过程主要由气温的降低启动，而在夏季光周期较短的南方，叶衰老过程主要由光周期的缩短启动，这一结果得到了大量野外和室内实验的支持。