基于气象因子的森林火灾面积预测模型_曲智林.pdfVIP

2007 12 18 12 Chinese Journal ofApplied Ecology, Dec. 2007, 18( 12): 2705- 2709 * * * 曲智林 胡海清 (, 150040) 通过统计分析理论研究了黑龙江省林火发生规律, 并建立了基于 象因子的森林火 灾面积预测模型. 结果表明: 兴安落叶松林区林火主要发生在 4) 6月和 10 月, 阔叶红松林区 林火主要发生在 3) 6月和 10月; 利用林火发生当日的平均风速相对湿度和平均温度的取 值范围可知, 兴安落叶松林区发生高等级林火概率较大的月份依次为 4月5月和 6月, 阔叶 红松林区则依次为 5月4月和 3月. 所建模型的平均精度达到 6313% , 能够较精确地预测林 火发生后林地可能的过火面积. 多元回归分析 象因子 林火面积 预测模型 1001-9332( 2007) 12-2705-05 S762. 1 A A prediction model for forest fire-burnt area based on m eteorological factors. QU Zhi-lin, HU H ai-qing (NortheastForestry University, Harbin 150040, China). -Chin. J. Appl. Ecol. , 2007, 18( 12): 2705- 2709. Abstract: ased on statistic analysis theory, the occurrence patterns of forest fire in Heilongjiang Provincewere studied, and the predictionmodel for forest fire-burnt areawas established based on meteorological factors. The results showed thatmost of forest fires inLarix g elinii forest region oc- curred from April to June and in October, and those in broadleavedKorean pine forest regionmainly occurred from M arch to June and in October. y adopting the values of averagew ind speed, relative hum idity and mean temperature, itwas predicted that in L. g elinii forest region, the first three monthsw ith greater probability of larger fire occurrencewereMarch, M ay and June in order, while in broadleavedKorean pine forest region, they wereM ay, M arch and Apri.l The average precision of themodelwas 6313%, suggesting that it could be used to predict the burnt area by forest fires. Key words: mult-i regression analysis; meteorological factor; burnt area; pre