基于Logistic回归的林缘计划烧除气象条件判别分析

摘要

森林火灾是一种破坏性极强的灾害,它与农事活动以及气象条件密切相关,加强其预防与控制具有十分重要的作用和意义.根据2014年~2015年闽东屏南县进行的60次林缘计划烧除试验观测的主要气象要素数据(天晴日数、温度、湿度、风速),建立基于 Logistic二元回归的林缘烧除气象条件判别模型进行检验和分析.结果表明,模型判别计划烧除成功与否的预报准确率高(91.7%),主要气象要素对林缘计划烧除成功与否发挥重要作用,其中连续晴天日数为最大影响因子,其次为风速.模型通过检验和验证结果显著,具有预报服务应用价值.%Forest fire is closely related to farming activities and meteorological conditions.A discriminant model of meteorological conditions is developed based on observation data of main meteorological factors (sunny days,temperature, humidity and wind speed)obtained in 60 times of plan burning tests in the forest edge of Pingnan county in Eastern Fujian Province from 2014 to 2015.The model is verified and analyzed by binary Logistic regression.The results show that,the model has high accuracy (91.7%)of judging the success of planning burning,and the main meteorological factors play an important role in success of burning plan in forest edge.The most significant impact factor is continuous sunny days,and also wind speed has important effect.