应用前期ENSO指标做棉铃虫大发生预测被引量：11

Using advance ENSO indices to forecast outbreaks of Helicoverpa armigera

下载PDF

导出

摘要分析了山东郓城26a(1974～1999)和德州22a(1978～1999)棉铃虫3代百株累计卵量、江苏丰县20a(1980～1999)棉铃虫2代百株累计卵量与从前两年1月份开始到当年7月份的ENSO指标(包括厄尔尼诺5个海温区N12、N3、N4、NC、NW的月平均海温距平)和南方涛动指数(SOI)的遥相关关系。遥相关分析结果表明德州、郓城三代卵量和丰县二代卵量与ENSO各指标遥相关关系的时间变化规律很相似,与各时段的N4均呈正相关,与NW和SOI大多数月份呈负相关。从中筛选出相关显著(p<0.05)的区域和时段作为预测因子,根据判别分析法用不同因子或因子组合分别建立了郓城、德州棉铃虫三代卵、丰县棉铃虫二代卵量的大发生预测模型,并对各模型进行回测检验及5～6a的预测检验,根据其预测效果筛选出最佳的灾变预测模型。结果表明,N4区的因子或因子组合建立的模型对德州、郓城棉铃虫三代卵量和丰县棉铃虫二代卵量的预测效果最好,可提前15～25个月做出大发生预测。 Statistical evidences is presented to show a teleconnection between population density (egg counts) of cotton bollworm (CBW), Helicoverpa armigera Hübner, and ENSO indices during the previous 2.5 years (from January of two years before to the current July). The ENSO indices used were the Southern Oscillation Index (SOI) and the Sea Surface Temperature (SST) anomalies in the five El Nio regions: Nio1&2 region (N12, 0～10°S,90～80°W), Nio3 region (N3, 5°N～5°S,150°～90°W ), Nio4 region (N4, 5°N～5°S, 160°E～150°W), NioCentral region (NC, 0°～10°S, 90°～180°W) and NioWestern region (NW, 0°～15°N, 130°～150°E). These ENSO indices were correlated with CBW survey data over 26 years (1974～1999) in Yuncheng, 22 years (1978～1999) in Dezhou, Shandong Province and 20 years (1980～1999) in Fengxian, Jiangsu province. We found significant (p<005) or highly significant (p<001) correlations between changes in CBW populations and ENSO indices. The abundance of CBW eggs was positively correlated with the SST anomaly in the N4 region and negatively correlated with the SST anomaly in NW region and with the SOI. Third generation eggs in Yuncheng and second generation eggs in Fengxian were significantly (p<005) correlated with ENSO indices during some months for all regions except N12. The only significant (p<005) correlation of egg numbers with SOI was for October of previous year, and the earliest significant (p<005) correlation with SST in the NW region was December of two years ago. The earliest month that showed significant (p<005) correlation between the third generation eggs of CBW in Dezhou and the SST anomaly in the NW region was November of two years ago, and was April of two years ago with SOI. The earliest significant (p<005) correlation for the third generation eggs in Yuncheng and Dezhou was with the SST of April of two years ago in N4, and was with the SST of August of two years ago for the second generation eggs in Fengxian. There was no significant (p>005) correlation between the 3rd generation eggs in Dezhou and N12, N3 and NC regions. The above significant SST anomalies were used as key factors to build predictive models for the forecasting of outbreaks of CBW by discriminant analysis. An outbreak was indicated if egg density exceeded the levels used by NATESC for population density of H. armigera in China: 3rd generation eggs were >85 in Yuncheng and Dezhou, or 2nd generation eggs were >500 in Fengxian. Outbreaks were designated as level 1, while lower nonoutbreak densities were level 0 for the purpose of this analysis. Several predictive models for longterm forecasting of the amount of CBW eggs were established using data up to 1994. These were tested for predictive accuracy against 1995～1999 CBW data. The results suggested that the SST anomaly in Nio4 region as the most important factor for forecasting CBW eggs in Fengxian, Yuncheng and Dezhou, but other ENSO indices including SOI showed poor predictive ability. The models are able to make forecasts 15~25 months ahead and had a 70% agreement with historical data (pre 1995) and had a predictive accuracy with 1995~1999 data of 78% (25 of 32 forecasts correct). The possible mechanism for the influence of ENSO on outbreak of H. armigera in China is discussed.