华北平原冬小麦对未来极端气候条件的响应及适应模式优化研究

The North China Plain (NCP) is an important granary of crop production in China. Extreme climate events have occurred frequently in the NCP, especially during the growth period of winter wheat, and these have significantly negative effects on winter wheat growth, development and ultimate yield. Under the background of climate warming, the frequency and intensity of extreme climate events in the future will continue to rise, and the natural risks and instabilities of agricultural production will become more prominent. Accurately predicting the occurrence of extreme climate events during the growth period of winter wheat and its impact on winter wheat production in the future are an important issue and challenge for the plain. By using multiple extreme climate indices that can represent temperature and water stress of winter wheat, this study will assess the trend of extreme climate events during the growth period of winter wheat based on the future climate scenario data. Based on the machine learning method, the effect of extreme climate on winter wheat yield is combined with the simulation results of crop model, which can made up the deficiency of current crop model that could not fully consider the effect of extreme climate conditions on crop growth and yield. With the crop model based on machine learning，this study will estimate the effect of future extreme climate on winter wheat production and explore the optimal management measures for winter wheat production in the future. The results of this study have significant meanings for implementing sustainable development of regional agriculture and ensuring food security, and will provide the critical theoretical basis for decision-makers in responding to future climate change and extreme climate events.

我国重要粮食生产基地华北平原极端气候事件频繁发生，尤其在冬小麦生育期更为明显，对冬小麦生产带来显著的负面效应。气候变暖背景下，未来极端气候事件发生的频次和强度也将持续上升，农业生产的自然风险和不稳定性将更为突出。准确预估未来冬小麦生育期内极端气候事件的发生及其对冬小麦生产的影响是该区域面临的重要课题和挑战。本项目选择能够表达冬小麦受温度和水分胁迫程度的多个极端气候指数，基于未来气候情景数据，评估冬小麦生育期内极端气候事件的变化趋势；基于机器学习法，将极端气候对冬小麦产量的影响结合到作物机理模型模拟结果中，弥补当前作物机理模型不能充分考虑极端气候条件对作物生长影响的不足；利用基于机器学习的作物机理模型预估未来极端气候对冬小麦生产的影响，并探索未来冬小麦生产的最优管理模式。本项目研究结果对维持区域农业可持续性发展，保证粮食安全具有重要意义，为应对未来气候变化和极端气候事件提供决策依据和理论。

我国重要粮食生产基地华北平原极端气候事件频繁发生，对冬小麦生产带来显著的负面效应。准确预估未来冬小麦生育期内极端气候事件的发生及其对冬小麦生产的影响是该区域面临的重要挑战。本项目选择能够表达冬小麦受温度和水分胁迫程度的极端气候指数，基于未来气候情景数据，评估冬小麦生育期内极端气候事件的变化趋势；基于机器学习法，将极端气候对冬小麦产量的影响结合到作物机理模型模拟结果中，改进作物模型对极端气候条件下作物生长的模拟能力；利用基于机器学习的作物机理模型预估未来极端气候对冬小麦生产的影响，并探索未来冬小麦生产的优化管理模式。研究发现未来华北平原冬小麦种植带的极端高温事件发生的频率和强度将增加，但极端低温事件将减少。极端降水事件无显著性变化，但具有很强的空间差异性。总体来说，华北平原未来冬小麦生产可能会受到更多极端气候事件的胁迫，特别是高温胁迫。利用机器学习算法对APSIM模型的输出结果进行额外的校正可以有效提高作物模拟的准确性。单独使用APSIM模型和基于机器学习的APSIM+RF-GA模型预估的未来华北平原冬小麦产量均增加，但单独使用APSIM模型与APSIM+RF-GA模型相比高估了未来冬小麦产量1–10%。华北平原除南部地区冬小麦为雨养，其它地区冬小麦的最佳灌溉次数是3次共240mm/ha。冬小麦的最优施氮量为210kg/ha。 受气候条件和品种特性影响，不同地区冬小麦的最佳播种窗口不同。在未来气候情景下弱春性冬小麦品种在华北平原具有更好的产量表现。相对于传统管理，优化管理能以较低的环境成本保证农业系统的产量。我们的研究结果为提高作物模型在极端气候条件下的模拟能力提供了一种新的思路和方法。此外，本研究还展示了一种利用作物模型设计可持续集约化耕作系统的有效方法，该方法可适用于其他耕作系统和地区。