欢迎访问林业科学,今天是

林业科学 ›› 2021, Vol. 57 ›› Issue (7): 121-130.doi: 10.11707/j.1001-7488.20210713

• 论文与研究报告 • 上一篇    下一篇

基于Logistic回归的兴安落叶松林飞火引燃试验

宁吉彬,耿道通,于宏洲,邸雪颖,杨光*   

  1. 东北林业大学林学院 森林生态系统可持续经营教育部重点实验室 哈尔滨 150040
  • 收稿日期:2019-11-07 出版日期:2021-07-25 发布日期:2021-09-02
  • 通讯作者: 杨光
  • 基金资助:
    "十三五"国家重点研发计划子课题(2017YFD0600106-2);国家自然科学基金项目(31400551);国家自然科学基金项目(31870644);国家自然科学基金项目(31700575)

Experiment on Spotting Ignition of Larix gmelinii Forest Based on Logistic Regression

Jibin Ning,Daotong Geng,Hongzhou Yu,Xueying Di,Guang Yang*   

  1. School of Forestry, Northeast Forestry University Key Laboratory of Sustainable Forest Ecosystem Management of Ministry of Education Harbin 150040
  • Received:2019-11-07 Online:2021-07-25 Published:2021-09-02
  • Contact: Guang Yang

摘要:

目的: 基于室内模拟的飞火引燃试验数据,使用Logistic模型对影响飞火引燃的因子进行建模,探究该模型在预测飞火引燃中的适用性,为林火行为研究提供依据和参考。方法: 选取黑龙江大兴安岭兴安落叶松纯林为对象,以球果、1 h时滞、10 h时滞小枝为火源,构造不同风速、可燃物含水率和压缩比的可燃物床层,进行引燃试验,建立以Logistic模型为基础的3种火源的引燃概率模型。结果: 3种类型火源分别进行了1 800次点烧试验,球果、1 h时滞和10 h时滞小枝引燃次数分别为414、161和337次;本研究梯度范围内,火源引燃概率随着风速增加而显著增加;可燃物含水率与引燃概率负相关,但受火源的干质量影响较大,在可燃物含水率为40%时,较大的3种火源干质量造成了引燃率升高;床层压缩比对火源引燃概率并没有表现出明显的正或负相关性;构建的球果模型预测引燃准确率为87.2%,总预测准确率为71.1%;1 h时滞小枝模型预测引燃准确率为79.6%,总预测准确率为78.6%;10 h时滞小枝模型预测引燃准确率为81.5%,总预测准确率为79.5%。结论: 3种火源引燃能力为球果>10 h时滞小枝>1 h时滞小枝,风速、可燃物含水率和火源干质量对引燃有较大作用,建立的3种Logistic引燃概率模型有较高准确率,可供飞火引燃预测时参考,以提高森林火灾扑救效率和减少扑救伤亡。

关键词: 兴安落叶松, 飞火, 火源, 风速, 含水率, Logistic

Abstract:

Objective: Based on the laboratory simulated experimental data of spotting ignition,Logistic model was used to establish a model of factors influencing the spotting fire ignition,and the applicability of Logistic model in predicting spotting ignition was explored,in order to understand the formation mechanism of spotting fire and provide reference to fire behavior forecast. Method: In this study,Larix gmelinii of Daxing'an mountains,Heilongjiang province was targeted. The cones,1 hour and 10 hours time-lag twigs were used as firebrands. Fuel beds with different wind speed,packing ratios and moisture content of combustible substance were constructed for the ignited experiment. Three Logistic models were established for predicting ignited probability of 3 spotting firebrands. Result: A total of 1 800 ignited experiments were conducted for each firebrand,cone 414 times,1 hour time-lag twig 161 times,and 10 hours time-lag twig 337 times,respectively. In the range of experiment designed,wind speed was positively correlated with ignited probability; fuel moisture content was negatively correlated with ignited probability,but the oven-dry weight of fire firebrand had a great influence on ignited probability. When fuel moisture content was 40%,the larger oven-dry weight of 3 firebrands caused the increase of ignition rate. There was no significant correlation between packing ratio and ignited probability. In cone firebrand model,predicting ignition accuracy rate was 87.2%,total predicted accuracy rate was 71.1%. In 1 hour time-lag twig firebrand model,predicting ignition accuracy rate was 79.6%,total predicted accuracy rate was 78.6%. In 10 hour time-lag twig firebrand mode model,predicting ignition accuracy rate was 81.5%,total predicted accuracy rate was 79.5%. Conclusion: The ignition capability of 3 firebrands is in the order of cone,10-hour time lag twig and 1-hour time lag twig. Wind speed and fuel moisture content and firebrand oven-dry weight have significant effect on ignition. All 3 ignited probability Logistic models have high accuracy rate,and they can be used as a reference for the prediction of spotting fire,in order to improve the efficiency of forest fire fighting and reduce casualties.

Key words: Larix gmelinii, spotting fire, firebrand, wind speed, moisture content, Logistic

中图分类号: