摘要
利用广州市近20年的森林火灾数据，以受害率、发生率、控制率及节气火灾发生概率构造时间序列，借助离散正交小波及Mann—Kendall方法分析林火发生时间序列特征，并结合博克斯-詹金斯法、马尔可夫链与聚类分析、判别分析进行短期预测．结果表明：林火发生时间序列有明显的波动和平稳的时间区间，年控制率从一种状态到另一种状态的飞跃点为1986年t在未来的两年里，火灾发生的状况将保持稳定，与1990～1998年和2001-2006年的火灾发生状况处于相同水平；林火的节气发生概率较高为立春、大雪2个节气时间段，防火期外和防火期内的白露和秋分两个节气区间段林火发生的概率较低；阳历新年和阴历新年前后林火发生概率高，呈现逐年增加的趋势．
A time series including victimized rate, incidence rate, control rate and the probability of solar-term forest fire was constructed based on 20 years＇ forest-fire data in Guangzhou City. A discriminate forecast of the future status of its forest-fire hazards was made according to a hazard analysis of the time-series features conducted by means of Discrete Orthogonal Wavelets, Mann- Kendall, Box-Jenkins, and Markov Chain coupling with Cluster Analysis. The results show that this time series has its own intervals for stability and fluctuation, that the control rate has a leap point from one state to another in 1986; that the next two years will see the state of fire occurrence remain stable and the fire situation in the same level as in the years of 1990-1998 and 2001-2006; and that forest-fire probability is very high on the solar-terms of Spring Beginning and Great Snow and still has an increasing tendency around Gregorian and Lunar New Year.
出处
《中南林业科技大学学报》
CAS
CSCD
北大核心
2008年第6期1-8,共8页
Journal of Central South University of Forestry & Technology
基金
国家林业局科技项目"长江流域退化天然林改造技术研究"(编号:069320)
关键词
森林保护学
森林火灾
时间序列
广州市
forest protection
forest fire
time series
Guangzhou City
