中南林业科技大学学报

利用 2005—2011年 5—10月黑龙江省大兴安岭林区的雷击火灾、闪电气象数据和其它数据，建立用于该地区雷击火预报的二项 Logistic回归模型，并将 ROC曲线用于模型的评价。结果表明： 2005—2011年该地区雷击火发生数量年际变化较大，相对于人为火呈上升趋势；集中发生于 5—6月，与云地闪电发生频数相关不显著 (P＞ 0.05)。经似然比检验，最终进入模型的变量为：前 7天 FFMC平均值、前 3天地闪次数、前 7天平均降水量。用于模型评价的 ROC曲线比较凸近左上角，曲线下面积 (AUC)为 0.888，表明模型预测效果较理想，可为大兴安岭林区的雷击火预测提供科学参考。

This paper used lightning fire data, lightning data, meteorology data and other useful data from May to October during 2005 and 2011 of Daxing’an ling Region in Heilongjiang Province, establishing Binary Logistic Regression Model to predict lightning fire in this region, and using ROC curve to evaluate the model. The results showed that lightning fires in this area had a great interannual variability between 2005 and 2011, and showing an increasing trend compared with human caused fires; mainly occurred in May and June, and not significantly correlate with the number of lightning(P＞ 0.05). Through likelihood ratio testing, the final variables using in the model were: previous 7 days’ average FFMC; previous 3 days’ average lightning frequency; previous 7 days’ average precipitation. The ROC curve which was used to evaluate the model was closer to the top left corner, the value of the area under the curve (AUC) was 0.888, indicating that the result of prediction simulation was desirable, and can be used to provide scientific reference for the prediction of lightning fires in Daxing’an ling Region.

引言
2 研究方法
3 结果与分析
4 结论与讨论

图1 雷击火与人为火年分布 <br/>Fig.1 Yearly distribution of lightning-caused and human-caused forest fires

图1 雷击火与人为火年分布
Fig.1 Yearly distribution of lightning-caused and human-caused forest fires

图2 2005—2011年大兴安岭地区雷击火分布 <br/>Fig.2 The distribution of lightning fires in Daxing’anling region from 2005 to 2011

图2 2005—2011年大兴安岭地区雷击火分布
Fig.2 The distribution of lightning fires in Daxing’anling region from 2005 to 2011

图3 2005—2011年雷击火次数和闪电次数的变化情况 <br/>Fig.3 Interannual dynamics of number of lightning fires and lightning from 2005 to 2011

图3 2005—2011年雷击火次数和闪电次数的变化情况
Fig.3 Interannual dynamics of number of lightning fires and lightning from 2005 to 2011

图4 5—9月雷击火次数和闪电次数的变化情况 <br/>Fig.4 Monthly dynamics of the number of lightning fires and lightning from March to September

图4 5—9月雷击火次数和闪电次数的变化情况
Fig.4 Monthly dynamics of the number of lightning fires and lightning from March to September

表1 所有变量显著性得分表<br/>Table 1 Sores of significance

表1 所有变量显著性得分表
Table 1 Sores of significance

表2 最后Logistic 回归结果 <br/>Table 2 The final results of Logistic regression

表2 最后Logistic 回归结果
Table 2 The final results of Logistic regression

图5 变量 pre_1的 ROC曲线 <br/>Fig.5 The ROC curve of the variable pre_1

图5 变量 pre_1的 ROC曲线
Fig.5 The ROC curve of the variable pre_1

表3 变量pre_1的曲线下面积 <br/>Table 3 Area under the curve of the variable pre_1 and variable pre_2

表3 变量pre_1的曲线下面积
Table 3 Area under the curve of the variable pre_1 and variable pre_2

[1] Wotton B M, Martell D L. A lightning fire occurrence model for Ontario[J]. Canadian Journal of Forest Research, 2005, 35: 1389- 1401
[2] Andrews P L. Fire danger rating and fire behavior prediction in the United States[J]. Proceedings of Fifth NRIFD Symposium International Symposium, Japan, 2005: 106-117.
[3] 田晓瑞 , 舒立福 , 赵凤君 , 等 . 大兴安岭雷击火发生条件分析 [J]. 林业科学 , 2012, 48(7): 98-103.
[4] 张吉利 , 毕武 , 王晓红 , 等 . 雷击火发生的影响因子与预测研究进展 [J]. 应用生态学报 , 2013, 24(9): 2674- 2684.
[5] Díaz-Avalos C, Peterson D L, Alvarado E, et al. Space time modelling of lightning-caused ignitions in the Blue Mountains[R]. Oregon: Canadian Journal of Forest Research, 2001: 1579- 1593.
[6] Latham D J, Schlieter J A. Ignition Probabilities of Wildland Fuels based on Simulated Lightning Discharges[R]. Washington: USDA, 1989. INT- 411.
[7] Larjavaara M, Kuuluvainen T, Rita H. Spatial distribution of lightning-ignited forest fires in Finland[J]. Forest Ecology and Management, 2005, 208: 177- 188.
[8] 曲智林 , 胡海清 . 基于气象因子的森林火灾面积预测模型 [J]. 应用生态学报 , 2007, 18(12): 2705- 2709.
[9] 金森 , 王晓红 , 于宏洲 . 林火行为预测和森林火险预报中气象场的插值方法 [J]. 中南林业科技大学学报 , 2012,32(6):1- 7.
[10] 王晓红 ,张吉利 ,金森 . 林火蔓延模拟的研究进展 [J]. 中南林业科技大学学报 , 2013, 33(10): 69-78.
[11] 王明玉 ,李华 ,舒立福 , 等 . 不同植被类型森林火灾及雷击火自组织临界性 [J]. 生态学报 , 2004, 24(8): 1803- 1807.
[12] 雷小丽 ,周广胜 ,贾丙瑞 , 等 . 大兴安岭地区森林雷击火与闪电的关系 [J]. 应用生态学报 , 2012, 23(7): 1743- 1750.
[13] 贾丙瑞 , 周广胜 , 于文颖 , 等 . 1972—2005年大兴安岭林区雷击火特征及其与干旱指数的关系 [J]. 林业科学 , 2011, 47(11): 99- 105.
[14] Ordóñez C, Saavedra A, Rodríguez-PérezJ R, et al. Using model-based geostatistics to predict lightning-caused wildfires[J]. Environmental Modelling & Software, 2012, 29: 44- 50.
[15] Nieto H, Aguado I, García M, et al. Lightning-caused fires in central Spain: Development of a probability model of occurrence for two Spanish regions. Agricultural and Forest Meteorology, 2012, 162: 35-43
[16] Magnussen S, Taylor S W. Prediction of daily lightning-and human-caused fires in British Columbia[J]. International Journal of Wildland Fire, 2012, 21: 342- 356.
[17] 杨光 , 舒立福 , 邸雪颖 . 气候变化背景下黑龙江大兴安岭林区夏季火险变化趋势 [J]. 应用生态学报 , 2012, 23(11): 3157- 3163.
[18] Van Wagner C E, Pickett T L. Equations and FORTRAN program for the Canadian Forest Fire Weather Index System[R]. Ottawa: Canadian Forestry Service Technical Report, 1985.
[19] Feng Chen, Yongsheng Du, Shukui Niu, et al. Modeling Forest Lightning Fire Occurrence in the Daxinganling Mountains of Northeastern China with MAXENT[J]. Forests, 2015, 6(5): 14221438.
[20] 于成龙 , 刘丹 , 杜春英 , 等 . 大兴安岭地区雷击火发生的雷电条件分析 [J]. 西北农林科技大学学报 , 2010, 38(11): 64-70.

备注

引言

2 研究方法

3 结果与分析

4 结论与讨论

学报简介