【摘要】 以芜湖市1986～2020年水稻病虫害及水稻种植面积、逐月气象要素、国家气候中心逐月环流和海温指数数据，分析了各病虫害的变化特征和显著气象影响因子，并借助多元逐步回归方法，研究了一种基于环流和海温指数的皖江地区主要病虫害的长期定量预测方法。结果表明：芜湖市病害总体轻于虫害，病害中稻瘟病、纹枯病和稻曲病发生面积比显著增加，白叶枯显著减少；虫害中二化螟、稻纵卷叶螟和褐飞虱发生面积比显著增加，稻蝗显著减少。各病虫害均有显著相关的气象影响因子。在此基础上，借助多元逐步回归方法，利用环流、海温指数对气象要素的影响及其影响的滞后性，建立了基于这些指数的皖江地区病虫害的长期定量预测模型。模型建立前先遴选出与本地病虫害显著相关的环流、海温指数，然后将其依次放入多元逐步回归模型，最终建立了8种病虫害的预测模型。模型和公式中的自变量均通过了显著性检验，模型具有较好的拟合效果。预报效果验证结果表明，模型具有一定的预报能力，能够在年初定量预测本年度主要病虫害的发生面积比，为皖江地区农业气象服务的开展提供了技术支撑。更多还原

【Abstract】 Based on the data of rice pest and diseases, rice planting area, monthly meteorological elements, monthly circulation and SST indexes during 1986～2020, the variation characteristics and significant meteorological influencing factors of each pest and diseases were analyzed, and long-term prediction method of pest and diseases in Wanjiang area was studied by means of circulation and SST index and multiple stepwise regression method. The results showed that diseases in Wuhu were generally lighter than pest. In these diseases, the area ratio of blast disease, sheath blight disease and rice false smut increased significantly, and the area ratio of bacterial leaf blight decreased significantly. In these pests, the area ratio of chilo suppressalis,rice leaf roller and brown plant hopper increased significantly, and the area ratio of rice grasshopper decreased significantly.All the pests and diseases had significant meteorological influencing factors. On this basis, by the impact of circulation and SST indexes on the meteorological elements and lagging characteristic of that impact, long-term quantitative prediction models of rice pest and diseases in Wanjiang area were built by means of these indexes and multiple stepwise regression method. Before the models were built, the circulation and SST index which significantly correlated with local pest and diseases was selected, and then those indexes were inserted into the multiple stepwise regression models successively. The prediction models of 8 kinds of pest and diseases were built at last. The models and all the independent variables in them passed significance inspection, and also had better imitative effect. The validation results of prediction effect showed that these models had certain predictive ability,and could quantitatively predict annual occurrence area ratio of main pest and diseases at the beginning of the year. This Method can offer technical support for the agricultural meteorological services in Wanjiang area.更多还原