江西省早稻雨洗花灾害指标构建与灾损评估

构建早稻雨洗花灾害指标及适于早稻产量估算的灾损评估模型，对开展早稻雨洗花灾害监测、损失评估、灾害保险等具有重要意义。该文以江西省早稻为研究对象，利用1981-2015年14个水稻气象观测站逐日气象资料和农业气象观测资料，筛选出基于早稻抽穗扬花期间过程降水量、最大降水量、降水日数及实际产量的雨洗花灾害样本78个，在此基础上，利用相关分析、正态分布以及主成分回归法，建立了雨洗花灾害指标和灾损评估模型，并对其进行验证。结果表明:抽穗扬花期降水对雨洗花灾害形成有显著影响，其主要影响时段为抽穗扬花普遍期前后5 d内，关键时段为抽穗扬花普遍期前后3 d内。日降水量40 mm可作为早稻抽穗扬花期雨洗花灾害临界指标。以该指标为基础，统计日降水量不低于40 mm的降水日数及其对应的累积降水量，当累积降水量为40~170 mm时，为轻度雨洗花灾害，早稻一般减产小于15%，平均减产10%；当累积降水量不小于170 mm时，为重度雨洗花灾害，早稻一般减产不低于15%，平均减产22%。指标验证结果与历史实际灾害发生情况有较好的一致性。雨洗花灾损评估模型检验结果表明:雨洗花年模拟产量与实际产量吻合度较高，平均相对模拟误差为4.3%，78.0%的资料相对误差在5%以内，可利用该模型对雨洗花年的早稻减产率进行模拟和预测。

Index and Loss Estimation of Rain Washing Damage to Early Rice Pollen in Jiangxi Province

Rain washing damage to pollen is one of the main agrometeorological disasters of early rice in Jiangxi Province. However, there are few studies on the disaster index and loss estimation model of this disaster. And in routine agrometeorological service, there are no definite and targeted criterion and loss assessment basis of rain washing damage to pollen. Therefore, studies on disaster index and loss estimation mode of rain washing damage to pollen are of great importance to the disaster monitoring, loss assessment and agricultural disasters'' insurance management of early rice. Taking the disaster of rain washing damage to pollen in Jiangxi Province as research object, 78 disaster samples of rain washing damage to pollen are picked out based on analysis of long-term (1981-2015) meteorological conditions during the whole growth period of early rice in 14 agrometeorological stations, and historical data about the observation of agrometeorological disasters, diseases and insect pests. Afterwards, index and loss estimation model of rain washing damage to early rice pollen are determined based on correlation analysis, normal distribution and principal component regression method, and verified by independent samples. Results show that the rainfall during heading-flowering stage of early rice has a significant effect on the formation of rain washing damage to early rice pollen. Main and key influence periods are 5 and 3 days before and after the heading-flowering stage, respectively. The daily precipitation 40 mm can be used as the threshold for rain washing damage to pollen in heading-flowering stage of early rice. Based on this index, the number of days with total precipitation exceeding 40 mm and their corresponding accumulative precipitation are counted. When the accumulative precipitation is between 40 mm and 170 mm (light disaster), the yield reduction rate of early rice is generally less than 15%, and the average reduction rate is 10%. When the accumulative precipitation exceeds 170 mm (severe disaster), the yield reduction rate is generally more than 15%, and the average reduction rate is 22%. The grading indexes are detected to be basically consistent with the historical occurrence levels of rain washing damage to early rice pollen. And simulation results of loss estimation model show that simulated early rice yields are highly accordant with the actual yields, the average relative error is 4.3%, and the relative error of 78% data is within 5%. It indicates that the model can be used to simulate and predict the yield reduction rate of early rice when rain washing damages rice pollen.