基于零膨胀负二项模型对大兴安岭地区夏季火发生预报模型的研究

ZINB-based Research on a Model for Forecasting Occurrence of Forest Fires in Daxing'anling Area in Summer

大兴安岭林区地处寒温带,是我国东北最重要的森林资源采伐基地。在全球气候逐渐变暖背景下大兴安岭地区的气候发生了明显的变化,导致夏季火频发,对该地区生态系统造成一定影响。选用零膨胀负二项(ZINB)模型,使用R-project软件,对该地区1967-2008年夏季火每日发生林火次数与气象因子之间的关系进行拟合,并用检验数据进行模型准确度预测。结果表明:最高气温对于模型的点部分影响极显著,平均地温和日照时数对于模型的零膨胀部分具有显著性影响,模型预报的准确率达71%。该模型对大兴安岭地区夏季火的预测准确性较好,利用该模型的预测预报结果能够为森林防火工作者合理配置灭火人力与物力,优先对火灾潜在高发区的可燃物进行中断管理提供科学的依据与技术支撑。

Daxing’anling forest area,with cold temperate climate,is the most important forest resource harvesting base in northeast China. Against the background of gradual global warming,the climate in Daxing’anling forest area has changed significantly,resulting in the frequent occurrence of forest fires in summer,which has generated certain effect on the ecosystem in this area. Zero-inflation negative binomial（ ZINB） selected,R-project software used,the fitting of the relationship between the number of daily forest fires and the meteorological factors in summers from 1967 to 2008 is conducted,with validation data used for model accuracy degree forecast. The result shows that the maximum temperature has a significant effect on the point part of the model,and that the average ground temperature and the number of sunshine hours have a significant effect on the zero-inflation part,with the accuracy rate of model forecastreaching up to 71%. This model shows good accuracy in forecast of summer forest fires in Daxing’anling area,and the forecast result of this model can provide scientific basis and technical support for forest fire workers’ rational allocation of firefighting manpower and material resources and priority given to interruption management of combustibles in areas involving potentially high fire risks.