Modeling Probability of Ignition in Taiwan Red Pine Forests

The main purpose of this study is to model the relationship between simulated human-caused fires and biophysical variables related to meteorological factors and fuel properties in Taiwan red pine forests. The experiment was carried out from August to December 1998 in the Dajashi National Forest. Three to 5 days were randomly selected to conduct the experiment monthly. Ignitions were performed hourly by igniting wooden matches and dropping them simultaneously onto a fuel bed hourly within the period from 9 a.m. to 4 p.m. on each sampling day. A logistic model was chosen to analyze the tests. One hundred and nine trials were conducted, and 46 of these trials were successful ignitions. The univariate and multivariate regression analyses were used respectively to fit the model. Results show that the best individual predictors were moisture content of pine needles (R2 = 0.83) and relative humidity (R2 = 0.82) in univariate regression analysis. Three variables, fuel moisture content of pine needles, wind speed, and fuel shading, fit the multivariate model (R2 = 0.93). Results indicate that the equations can be used to help predict fire danger in Taiwan red pine forests.

本研究的核心目的为构建台湾红松林中人为模拟野火与气象因子、可燃物特性相关生物物理变量之间的关系模型。实验于1998年8月至12月在达加什国家森林开展，每月随机选取3至5天进行实验。每个采样日的上午9时至下午4时期间，每小时通过点燃木质火柴并同时将其投放至可燃物床上来完成点火操作。本研究选用逻辑回归模型对实验数据进行分析，共计完成109组试验，其中46组成功实现点火。分别采用单变量回归分析与多变量回归分析方法对模型进行拟合。单变量回归分析结果显示，表现最优的单个预测因子为松针含水率（R²=0.83）与相对湿度（R²=0.82）；多变量模型选取松针可燃物含水率、风速以及可燃物遮阴度三个变量，拟合得到的决定系数为R²=0.93。研究结果表明，该模型方程可用于辅助预测台湾红松林的火险等级。