多尺度、多灾种和多过程下华北平原冬小麦灾害损失风险评估研究

Natural hazards usually occur randomly and occasionally. The information during these dynamic courses is obtained difficultly, which have caused the basic theory and methodology in the field of disaster risk science are only focused on a single hazard, rather than multi-hazards. However, crop will encounter multi-natural hazards during the long period of growth and development. Therefore, traditional statistical method is not able to identify scientifically yield loss. Such weakness does not only hinder the development of disaster risk science, but also limit the methods and achievements of disaster risk sciences to be applied extensively by the related departments (e.g., “the uniform premium rating in a province”, “yield loss identified unknowingly by a definite disaster” etc., such problems encountered by crop insurance companies). Therefore, it is highly necessary and significant to perform the risk assessment of agricultural meteorological disasters (AMD) on multi-scales, multi-hazards, and multi-processes to understand the mechanisms controlling the final losses. Firstly, the developed MCWLA-Wheat model will be improved to strengthen its ability to simulate the impact of extreme events (drought, high temperature, dry-hot wind). Then, the improved model will be parameterized at the grid (0.25°×0.25°) scale, and be validated at the two scales (county and province). Then, by inputting the different scenarios of AMDs, the MCWLA-Wheat will simulate the dynamic processes responding to these stress scenarios, and the respective yield losses will be calculated by MCWLA-Wheat. The study will finally construct the models to catch the relationship between “a single\multi-hazards - yield loss risk” and assess the integrated risk for winter wheat in North China Plain. We are sure that the achievements in the study will provide new theory and method for the disaster risk field, and will benefit the related departments to assess rapidly the yield loss from AMDs, and to set a reasonable premium rating.

自然灾害是随机和偶然发生的。人们很难获取灾变过程的详尽资料，数据的缺乏导致了灾害风险学科的基本理论方法大都针对单灾种。但是作物整个生长过程可能遭遇多种自然灾害，因此传统的统计方法无法科学厘定农业产量的损失。这种不足既阻碍了自然灾害风险学科的发展，也使得该学科理论和成果的实际运用受到了限制（例如保险业的“全省统一费率”“产量损失的灾种归因不分”等）。因此有必要开展多尺度、多灾种和多过程下农业气象灾害风险研究以便揭示灾害发生的机理过程。本研究首先改进MCWLA-Wheat对干旱、高温、干热风的模拟能力，基于格点尺度（0.25°×0.25°）率定和校准模型；然后运行该模型实现不同灾害事件胁迫下作物生长发育过程的动态模拟及产量损失厘定； 最终实现多尺度下“单/多灾种-损失风险”建模和冬小麦综合风险评估。本项目将为灾害风险学科提供一套新的理论方法、也将有利于农业气象灾损的快速评估和保险费率的厘定。

针对目前我国保险行业的“全省统一费率”“产量损失的灾种归因不分”等困境，本研究开展了多尺度、多灾种和多过程下的农业气象灾害风险评估研究。首先我们通过结合遥感、优化、同化等技术改进了MCWLA-Wheat模型的模拟能力（R2 0.26提高到0.42；RMSE从1021 降低到737 kg/ha）；同时通过设置不同灾害情景驱动改进的作物模型，了解了干旱、高温、干热风对作物关键生长过程的影响；并在三个空间尺度（0.25°×0.25°格点、县级、省级）完成了灾害的快速评估。我们开发的这套系统能实现不同灾害事件胁迫下作物生长发育的动态模拟及产量的损失评估，并在多尺度下 构建“单/多灾种-损失风险” 模型，完成大范围农作物产量损失的客观定损和核损。本研究的成果为防灾减灾政策制定、农业气象灾情的快速评估、保险公司费率厘定等提供理论支持和技术指导。