面向农业气象的区域地球系统模式技术路径与应用

Technical Pathways and Applications of Regional Earth System Models for Agricultural Meteorology

摘要:

在全球气候变化背景下,农业干旱、高温热害、低温冻害等极端气候事件频发,对我国农业生产稳定性与粮食安全保障构成严峻威胁。传统农业气象模型多采用单点、离线驱动的运行方式,难以表征农田下垫面与大气之间的动态双向反馈。相较于侧重大尺度气候长期演变的全球地球系统模式,区域地球系统模式凭借更高的空间分辨率和更完善的复杂地形处理能力,可更细致地描述局地中小尺度天气过程及农田下垫面对区域气象条件的影响,并为区域农业气象影响评估和业务应用提供模拟基础。现有研究表明,动态作物过程与陆—气系统在线耦合模拟能够加强农田下垫面反馈过程的表达,但不同模式和耦合方式的适用性取决于研究目标、时空尺度及观测约束;氮循环已由陆面模式内部生产力约束拓展至陆—气—化学多圈层耦合,磷循环区域在线耦合仍处于早期发展阶段。当前区域地球系统模式仍面临作物与管理参数区域化不足、极端损伤与恢复过程表征简化、多过程耦合误差传播及观测约束不足等问题。因此,本文围绕区域气候模式、陆面模式、作物模型与区域地球系统模式之间的功能联系,系统梳理农业过程进入区域地球系统模式的主要技术路径,重点总结作物生长—陆面水热—边界层反馈,以及播种、灌溉、施肥、水资源和氮—磷循环等关键过程耦合表达,并进一步归纳在极端与复合农业气象事件、典型农业区模拟以及农业气象预测与决策支持中应用进展。在此基础上,提出“最小充分反馈链”模式选型原则,并从高分辨率模拟、多源状态—参数同化、模块化耦合、物理信息人工智能及区域地球系统模式与农业气象业务协同等方面讨论未来研发方向,为区域地球系统模式支撑农业气象灾害评估与决策服务提供参考。

Abstract:

Under global climate change, frequent agricultural droughts, heat stress, cold extremes, and other climate-related hazards pose increasing challenges to agricultural production and food security. Traditional agricultural meteorological models generally rely on offline simulations driven by prescribed meteorological forcing, making it difficult to represent dynamic feedbacks between cropland surfaces and the atmosphere. Compared with global Earth system models that focused on long-term, large-scale climate evolution, regional Earth system models (RESMs) with relatively high spatial resolution and improved representation of complex terrain can provide more detailed simulations of local- to mesoscale weather processes and the effects of cropland surface conditions on regional meteorology, thereby providing a modeling basis for regional agrometeorological impact assessment and operational applications. Recent applications in simulating extreme and compound agricultural meteorological events, representative agricultural regions, and agricultural meteorological prediction and decision-support systems were further discussed. Existing studies indicated that dynamic crop parameterizations and their online coupling with land-atmosphere systems can improve the representation of agricultural feedback processes, while model applicability depended on research objectives, spatial and temporal scales, and observational constraints. Nitrogen cycling had gradually expanded from internal land-surface productivity regulation to coupled land-atmosphere-chemical processes, whereas regional online coupling of phosphorus cycling remained under development. Current challenges included insufficient regionalization of crop and management parameters, simplified representations of extreme-event impacts and recovery processes, uncertainties in multi-process interactions, and limited observational constraints. This review summarized the development of RESMs for agricultural meteorological applications, focusing on the relationships among regional climate models, land surface models, crop models, and regional Earth system models. The major pathways for incorporating agricultural processes into regional models were systematically reviewed, including crop growth processes, land surface water-energy exchanges, agricultural management practices, and biogeochemical cycles. Future developments should focus on high-resolution simulations, multi-source state and parameter assimilation, modular coupling approaches, physics-informed artificial intelligence, and enhanced integration of regional models with agricultural meteorological services. This review can provide a reference for advancing regional Earth system models in agricultural meteorological risk assessment and decision support.