灾变模型在马尾松毛虫幼虫发生量预报中的应用

  目的  提高马尾松毛虫Dendrolimus punctatus幼虫发生量预测预报结果的准确性。  方法  利用灰色灾变预测GM(1, 1)模型预测了安徽省潜山县1989-2016年马尾松毛虫越冬代、1代和2代严重发生的年份。  结果  马尾松毛虫越冬代虫口数的GM(1, 1)灾变预测模型为:${\hat z^{(1)}}(k + 1) = 9.58075{{\rm{e}}^{0.26933k}} - 8.58075$, 其中k为年序号, ${\hat z^{(1)}}(k + 1)$为灾变年序号。1代幼虫虫口数的GM(1, 1)灾变预测模型为:${\hat z^{(1)}}(k + 1) = 18.1818{{\rm{e}}^{0.24187k}} - 17.1818$。2代幼虫虫口数的GM(1, 1)灾变预测模型为:${\hat z^{(1)}}$(k+1)=20.123 7e0.197 58k-19.123 7。根据此模型求得已知年份的拟合值与观察值, 对两者差异进行t检验, 差异均不显著, 即拟合值与观察值间吻合度高, 各灾变年精度值平均为84.40%, 84.85%和84.08%, 总体平均精度依次为96.25%, 92.34%和94.09%, 模型精度高。由此推算未来时刻的预测值得到, 从2011年马尾松毛虫越冬代幼虫灾变年算起, 再过10 a即2021年为马尾松毛虫越冬代大发生年。从2011年马尾松毛虫1代幼虫灾变年算起, 再过11 a即2022年为马尾松毛虫1代幼虫大发生年。从2011年马尾松毛虫2代幼虫灾变年算起, 再过9 a即2020年为马尾松毛虫2代幼虫大发生年。  结论  灾变预测对马尾松毛虫幼虫发生量灾变的预报是一种较理想的预报方法。

A catastrophe model to forecast larvae occurrence of Dendrolimus punctatus

  Objective  The aim is to improve control of Dendrolimus punctatus larval occurrence, and to select a suitable prediction model by increasing prediction accuracy through a catastrophe prediction method.  Method  The prediction model GM(1, 1) for D. punctatus was used over 28 years from 1989 to 2016 in Qianshan County, Anhui Province.  Result  The GM(1, 1) cataclysmic prediction model for the overwintering generation of D. punctatus was as follows:${\hat z^{(1)}}$(k+1)=9.580 75e0.269 33k-8.580 75, where k was the annual serial number and ${\hat z^{(1)}}$(k+1) was the disaster year serial number. The GM(1, 1) cataclysmic prediction model for the number of larvae of the first generation was:${\hat z^{(1)}}$(k+1)=18.181 8e0.241 87k-17.181 8, and the second generation larval population was:${\hat z^{(1)}}$(k+1)=20.123 7e0.197 58k-19.123 7. According to this model, fitted and observed values of known years were obtained. Results of a t test for t_0.05 were not significant. The average annual accuracy of three disasters was 84.40%, 84.85%, and 84.08% with the total average accuracy of the first being 96.25%, of the second being 92.34%, and of the third being 94.09%. The predicted future times were as follows:for the wintering D. punctatus disaster of 2011, it would be another 10 years (2021) before the next overwintering. For the first generation D. punctatus larvae in the catastrophic year 2011, it would be another 11 years (2022), until the great occurrence of the first generation larvae. For the second generation D. punctatus larvae, the next disaster after 2011 would occur 9 years later (2020).  Conclusion  Thus, the catastrophe prediction method could be an ideal method for predicting the occurrence of larval cataclysm with D. punctatus.