基于动态模型和遥感数据的冬小麦晚霜冻害监测方法研究

The freezing disaster by late frost is one of the major agricultural meteorological disasters for winter wheat in China. Accurate monitoring of regional-scale freezing disaster by late frost of winter wheat has a great role on agricultural disaster prevention and mitigation. At present, two outstanding problems regarding late frosting disaster are only using single factor and lacking of high-precision monitoring model. In this research, our study will be conducted from the perspective of formation mechanism of freezing disaster by late frost. The following studies will be carried out: a) the identification method of winter wheat growth stages and spike differentiation stage will be derived using remotely sensed vegetation index and accumulated-temperature-based winter wheat growth model; b) the crop growth model and soil-vegetation-atmosphere-transfer model and remotely sensed data will be coupled to obtain sensitive plants and environmental parameters to freezing disaster by late frost; c) a monitoring indicators dataset will be established using the lowest leaf temperature and daily minimum temperature in the growth cone layer and a factor dataset will be built by considering a variety of factors including growth stage, temperature, humidity, plants, terrain and other factors, base on which a multi-factor freezing disaster monitoring model will be constructed; d) by simulating the interaction between the growth of winter wheat frost injury and the environment a mechanic monitoring model also will be built, and the temporal and spatial adaptability of the model will be evaluated and verified. The study achievements will be useful for the monitoring of agro-meteorological disasters.

晚霜冻害是影响我国冬小麦生产的重大农业气象灾害之一, 准确监测区域尺度冬小麦晚霜冻害对于农业防灾减灾具有重要意义。针对目前冬小麦晚霜冻害监测中存在的监测因子单一和缺乏高精度监测模型等突出问题，本项目从冬小麦晚霜冻害致灾机理出发，研究应用植被指数和生长积温模型识别冬小麦生育时期和穗分化期的方法；在通过融合遥感信息、作物生长模型和土壤-植被-大气传输模型获取植株和环境参数的基础上，以最低叶温和生长锥所在层最低气温构建监测指标集，以生育阶段、温度、湿度、苗情和地形等因子构建监测因子集；采用集合典型相关分析等方法，进行多因子的晚霜冻害监测经验模型的研发；基于动态模型模拟的数据分析冬小麦霜冻害与生长发育环境的关系，构建冬小麦晚霜冻害机理模型，并对模型的时间和空间适应性进行评价和验证。本项目的特色和创新是融合遥感信息和动态模型以期构建机理性较强的冬小麦晚霜冻害监测模型，为农业气象灾害的监测和评估服务。

晚霜冻害是影响我国冬小麦生产的重大农业气象灾害之一, 准确监测区域尺度冬小麦晚霜冻害对于农业防灾减灾具有重要意义。针对目前冬小麦晚霜冻害监测中存在的监测因子单一和缺乏高精度监测模型等突出问题，综合考虑穗分化期、气温、叶温等多种因子，构建融合遥感信息和动态模型的晚霜冻害监测模型，提高监测精度。本项目基于WheatGrow模型构建了冬小麦拔节至抽穗期的穗分化期识别模型；基于土壤-植被-大气传输模型SHAW进行麦田气温、叶温廓线模拟，获得麦田冠层垂直方向分层气温和叶温数据；引入集合卡尔曼滤波算法同化MODIS LST，并耦合WOFOST作物生长模型，获取植株和环境参数；以叶温、冠层气温以及其他环境参数等因子构建监测因子集，以冬小麦产量构成要素的亩穗数、穗粒数和千粒重为指标集，进行单因子典型相关分析，进而基于回归集合平均方法构建了多因子集合典型相关分析模型；集成各子模型，形成基于遥感信息和动态模型的冬小麦晚霜冻害监测模型。本项目旨在进一步提升冬小麦晚霜冻害遥感监测水平，为农业气象灾害的监测和评估服务。