Abstract: The correlations between the situation indicators (peak time of ingoing, last length of peak period, and moth quantity in peak period) of rice leaf roller in Yixing, Yandu and Jingjiang of Jiangsu Province and the grid monthly sea surface temperature (SST) of west Pacific were analyzed by statistical method, and the correlation maps were produced by using GRADS software. The regions in which the SST was significantly correlated with the situation indicators were identified, and the SST at these regions, which was processed by optimization correlation technique, was used as the predictor to set up the long-term models for predicting the situation indicators of rice leaf roller during its immigration period in the three regions. The results showed that the immigration time of rice leaf roller in each of the regions was highly correlated with the SST in that region, and the duration of immigration peak was well correlated with the SST of west Pacific. The correlations between moth amount and SST were significant and stable, and showed some seasonality. Model calibrations indicated that the agreements between outputs from all models and observations were statistically significant (α＝0.01), and model validations demonstrated the applicability of the models developed in this study in predicting the situation indicators of rice leaf roller. These models were capable of predicting the possible occurrence situation of rice leaf roller one to two months in advance, being of significance in the prevention and control of rice leaf roller, suitable management of rice production, reduction of pesticide pollution, and protection of environment.

Key words: riverine nitrogen flux, net anthropogenic nitrogen inputs, response relationships, influential factors.