中国媒介生物学及控制杂志 ›› 2020, Vol. 31 ›› Issue (1): 21-26.DOI: 10.11853/j.issn.1003.8280.2020.01.005

• 论著 • 上一篇    下一篇

差分自回归移动平均模型在蚊密度分布特征预测中的应用

运玲, 王福才, 张秋芬   

  1. 唐山市疾病预防控制中心消杀科, 河北 唐山 063000
  • 收稿日期:2019-08-25 出版日期:2020-02-20 发布日期:2020-02-20
  • 作者简介:运玲,女,副主任医师,主要从事病媒生物防制工作,Email:845689649@qq.com
  • 基金资助:
    河北省2017年度医学科学研究重点课题计划(20171386)

Application of autoregressive integrated moving average model in prediction of the distribution characteristics of mosquito density in Tangshan, Hebei province, China

YUN Ling, WANG Fu-cai, ZHANG Qiu-fen   

  1. Tangshan City Center for Disease Control and Prevention, Tangshan 063000, Hebei Province, China
  • Received:2019-08-25 Online:2020-02-20 Published:2020-02-20
  • Supported by:
    Supported by the Key Issues of Medical Science Research in Hebei Province (No. 20171386)

摘要: 目的 分析唐山市不同年份成蚊种群密度分布,探讨差分自回归移动平均(ARIMA)模型对未来蚊虫密度的预测。方法 选用诱蚊灯法监测唐山市市区2010-2018年不同生境各蚊种的成蚊密度,运用ARIMA模型对2010-2017年各月总蚊密度数据建立模型,预测2018年各月的总蚊密度,并根据2018年实际监测密度评估预测效果。结果 2010-2017年唐山市市区总蚊密度为2.14只/(灯·h),淡色库蚊为优势蚊种,占捕获总数的98.02%;在不同生境中,以牲畜棚的蚊密度最高,为3.65只/(灯·h),不同生境、不同蚊种构成比差异有统计学意义(χ2=249.177,P<0.001);8年中以2014年蚊密度最高为3.00只/(灯·h),2016年蚊密度最低为1.41只/(灯·h);从季节消长看,总蚊密度高峰出现在7月,呈单峰曲线,蚊密度为4.71只/(灯·h)。运用ARIMA模型对唐山市市区2010-2017年各月总蚊密度进行拟合,选取ARIMA(0,1,1)×(2,1,0)12作为最佳模型,残差序列为白噪声序列(Q=20.654,P=0.148),用此模型预测2018年的总蚊密度,实际值均落入预测值的95%可信区间内,预测的蚊密度季节消长趋势与实际值基本吻合,可用于中短期蚊密度预测。结论 通过分析唐山市的蚊种密度和季节消长特征,成功建立ARIMA模型,可较准确地预测未来蚊密度,从而有效预警蚊媒疾病的暴发和流行。

关键词: 蚊密度, 监测, 差分自回归移动平均模型

Abstract: Objective To analyze the population density distributions of adult mosquitoes in different years in Tangshan, Hebei province, China, and to predict future mosquito density by using the autoregressive integrated moving average (ARIMA) model. Methods The lamp trapping method was used to monitor the densities of various adult mosquito populations in different habitats of urban Tangshan from 2010 to 2018. The ARIMA model was established based on the data of the total densities of mosquitoes in each month from 2010 to 2017. Then this model was used to predict the total densities of mosquitoes in every month of 2018, and the prediction effect was evaluated according to the actual densities monitored in 2018. Results During 2010 to 2017, the total density of mosquitoes in urban Tangshan was 2.14 individuals per lamp hour, and Culex pipiens pallens was the dominant species, accounting for 98.02% of the total number of captured mosquitoes. Among different habitats, livestock sheds had the highest mosquito density of 3.65 individuals per lamp hour. There was a significant difference in the composition ratio of mosquito species between different habitats (χ2=249.177, P<0.001). During the eight years, the highest mosquito density (3.00 individuals per lamp hour) was observed in 2014, and the lowest density (1.41 individuals per lamp hour) in 2016. Seasonal fluctuations in the total density of mosquitoes showed a single-peak curve, with the peak at 4.71 individuals per lamp hour in July. An ARIMA model was used to fit the total densities of mosquitoes in urban Tangshan in every month from 2010 to 2017. ARIMA (0, 1, 1)×(2, 1, 0)12 was selected as the best model, and the residual sequence was a white noise sequence (Q=20.654, P=0.148). Then the model was used to predict the total densities of mosquitoes in 2018, and the actual values fell within the 95% confidence interval of the predicted values. The predicted seasonal fluctuations in mosquito density were quite close to the actual values, suggesting this model could be used to predict mosquito density in short and medium term. Conclusion By analyzing the densities of mosquito populations and seasonal fluctuations in Tangshan, the ARIMA model was successfully established to predict future mosquito density, so as to effectively warn the outbreak and epidemic of mosquito-borne diseases.

Key words: Mosquito density, Surveillance, Autoregressive integrated moving average model

中图分类号: