Teleconnection Between The Advance SST, ENSO Indices And Early-season Immigration Of Nilaparvata Lugens (St(?)l) And Its Implications For Medium-and Long-term Forecasting

Brown Planthopper (BPH), Nilaparvata lugens (Stal) is a long distance migratory rice pest in Asia, and also is an important rice pest in South China. Since 1970s, BPH outbreaks have taken place frequently in China. BPH population and outbreak frequency showed downtrend from 1992 to 2002 but uptrend again from 2003. The accurate and advance forecasting of BPH occurrence can provide information for its prevention. At present, BPH forecasting focus on the short and medium-term prediction which is widely used in practice for their veracity. No deep researches have been done in BPH long-term forecasting. The early-season immigration of BPH correlates closely to the subsequent population in China. In addition, the meteorological factors play an important role in the whole immigration. So, this paper presented some results of my exploratory research on using the advance sea surface temperature (SST) or ENSO indices as early predictive factors of early-season immigration of BPH.1. Teleconnection AnalysisBased on BPH light-trap data of 16 representative BPH monitoring stations in different rice areas of China as well as SST data of the Pacific Ocean, Indian Ocean and ENSO indices, the relationship between early-season immigration of BPH and advance SST or ENSO indices was analyzed. A sustainable and significant correlation SST area was defined as that the correlation occurred simultaneously at least on 5 grids and lasted no less than 3 months continuously. After the teleconnection analysis, 196 SST areas with 305 periods of time in the Pacific Ocean, 142 SST areas with 305 periods of time in the Indian Ocean and 255 significant ENSO indices were obtained.2. Screening out predictive factorsAfter calculating the averages of monthly mean SST anomaly (SSTA) in a sustainable and significant correlation SST area, the correlation coefficient between these averages and early-season immigration of BPH was studied. The stability analysis of the correlation coefficient was made in order to delete unstable correlated SST areas or periods of time. Finally, 115 SST areas with 168 periods of time in the Pacific Ocean, 101 SST areas with 127 periods of time in the Indian Ocean were left. Their spatio-temporal distribution showed as follows:In the Pacific Ocean, 50% of the sustainable and significant correlation SST areas distributed in Central Pacific (30°N-30°S) while 25% of the areas in the South (<30°S) and North (>30°N) Pacific, respectively. In the Indian Ocean, SST areas stably and significantly related with early-season immigration of BPH in the south of Yangtze River Valley mainly distributed in the South (41.6%) and North (40%) Indian Ocean, while those significantly related with early-season immigration of BPH in the Yangtze River Valley and its North mostly distributed in the South (53.6%) and Central (26.8%) Indian Ocean. More than 70% of significant ENSO indices were in N3, N4 or N3+4 regions. As the temporal distribution, the sustainable and significant correlation SST areas in the Indian and Pacific Ocean or significant ENSO indices were mainly one and two years ahead, accounting for 81.5-87%.Through comparing the consistency of the spatio-temporal distribution of the teleconnection, we found that: early-season immigration of BPH in the south of South China, the north of Nanling Mountain or Yangtze River Valley usually appeared a significantly positive relationship with SSTA in the middle-east Equatorial Pacific from previous winter to the subsequent spring. It was assumed that positive Equatorial Indian SSTA indicated a mass of early-season immigration of BPH in the Yangtze River Valley and its south except the Nanling Mountain. Early-season immigration of BPH often correlated with SSTA in each Nino region negatively before previous spring and positively after previous spring.3. Building up and verification of the forecasting modelsThe average SSTA of all grids in one significant SST area or significant ENSO index was used as predictive factor to build the forecasting models for early-season immigration of BPH. Firstly, the factors were combined according to the advance year and the SST area or ENSO index. The last three years' data were set aside for model verification. Early-season immigration of BPH was divided into 3 levels including small immigration (X|-+S). Models were made by stepwise regression analysis. The historical accordance and predictive accuracy were calculated and then used for screening out the models as the standard of more than 50% and 1/3, respectively. There were 306 forecasting models based on SSTA in the Pacific Ocean, 199 based on SST in the Indian Ocean and 44 based on ENSO indices in total. 51.6% of the Pacific SSTA predictive factors were in Central Pacific and 67.3% of the Indian SSTA predictive factors were in the South Indian. The ENSO indices predictive factors were mainly SSTA in N3 or N3+4 region and SOI, accounting for 93.2%. As temporal distribution, the factors were the same as that of the teleconnection which were mainly one and two years ahead with the proportion of more than 90.4. Integrated ForecastThere were numerous models for early-season immigration of BPH in a single station, so integrated forecasting models were made to be convenient for extending and applying. Firstly, the single models were eliminated with the standard of forecasting error higher than 0.5 levels. Secondly, the weight coefficient of each model was calculated from the historical accordance and predictive accuracy. Finally, three integrated forecasting models in each monitoring station were classified and built up according to their forecasting advance years, which were ultra-long-term (two years ahead), long-term (one year ahead) and medium-term (this year) integrated forecasting models. At last there were 38 forecasting models based on SSTA in the Pacific Ocean, 19 based on SSTA in the Indian Ocean and 12 based on ENSO indices in total. The results suggested that these models allowed the forecasting of early-season immigration of BPH to be brought forward about 2-27 months with a good predictive accuracy of 87.9% (211/ 240 of probability).5. The possible mechanism of the teleconnectionThe possible mechanism for the influence of SST in the Pacific Ocean on early-season immigration of BPH was studied. On the base of SST, BPH light trap, BPH field population and meteorological data in Tongzhou and Huai'an, the relationship between every two of them were analyzed. Then the teleconnection were compared. Positive SSTA in the middle-east Equatorial Pacific Ocean last winter affected the area index of Subtropical High this June, which influenced the precipitation of the last ten days in June, accordingly influencing early-season immigration of BPH in Tongzhou. Negative SSTA in the middle-east Equatorial Pacific Ocean spring and summer two years ahead affected the north boundary of Subtropical High this May, which changed the precipitation of the first ten days in July, accordingly influencing early-season immigration of BPH in Huai'an. Finally, the hypothesis of SST→General circulation (Subtropical High)→Climate(precipitation)→ early-season immigration of BPH was assumed.