模拟降雨下以小时为步长的崇礼区典型林分地表细小死可燃物含水率预测模型

Prediction model of surface dead fine fuel moisture content at onehour intervals in a typical stand under simulated rainfall in the Chongli District,Zhangjiakou City,China

地表细小死可燃物含水率是影响林火蔓延的重要指标,构建精准的崇礼区地表细小死可燃物含水率预测模型,有利于完善崇礼区森林火灾预报系统.以白桦(Betula platyphylla)、落叶松(Larix gmelinii)、慢生杨(Populus L.)和山杨(Pobulus davidiana)4种崇礼区典型林分为研究对象,通过野外模拟降雨试验,包括4个降雨梯度(降雨量分别为1、2、5、10 mm),连续监测降雨后不同时间间隔(2、4、6、8、10、12、24 h)的地表细小死可燃物含水率动态数据,同时收集林分内气象因子,建立了适用于降雨期的以小时为步长的地表细小死可燃物含水率预测模型.结果显示:模拟降雨期内,落叶松林分的地表细小死可燃物含水率范围为30.68%-194.87%,显著高于其他林分.4种林分的地表细小死可燃物初始含水率均超过100%,随着时间的推移,含水率逐渐下降,24 h后含水率在27.16%-45.82%范围.时间间隔对4种林分的地表细小死可燃物含水率影响最大,相对重要性在23.72%-55.06%范围.温度则只对山杨林分地表细小死可燃物含水率有较大影响,相对重要性在30%左右.所建立的12个预测模型中,参数均通过显著性检验,残差满足同方差性,模型可解释超过60%的地表细小死可燃物含水率变异.模型的平均绝对误差(MAE)和均方根误差(RMSE)分别为19.24%和23.27%.本研究表明崇礼区不同林分地表细小可燃物含水率存在显著差异,降雨要素是解释含水率变化的主导因子.(图5表6参45)

The surface dead fine fuel moisture content is an important index that affects the spread of forest fires.Establishing an accurate prediction model for this index could aid the prediction of forest fires in the Chongli District.In this study,four typical forests of Betula platyphylla,Larix gmelinii,Populus L.,and Pobulus davidiana in the Chongli District were investigated.The experiment included four rainfall gradients(1,2,5,and 10 mm),the dynamic data on surface dead fine fuel moisture content at different intervals(2,4,6,8,10,12,and 24 h)after rainfall were continuously monitored,and meteorological factors within the stand were collected.Thus,a prediction model was established in an hourly step.The surface dead fine fuel moisture content in L.gmelinii ranged from 30.68%to 194.87%,which was significantly higher than the other contents.The initial content in the simulated rainfall period was more than 100%,and it gradually decreased to 27.16%–45.82%after 24 h.The time interval had the greatest influence on the surface dead fine fuel moisture content in most stands,and its relative importance ranged from 23.72%to 55.06%.Temperature only had a significant effect in P.davidiana,and its relative importance was approximately 30%.The parameters of the 12 established prediction models were tested for significance.The residual met the homoskedasticity,and the model could explain more than 60%of the variation in moisture content.The mean absolute error and root mean square error of the model were 19.24%and 23.27%,respectively.The results showed significant differences in the surface dead fine fuel moisture content between the different forest stands in Chongli District,and rainfall was the dominant factor contributing to the variation in fuel moisture content.The prediction model of surface dead fine fuel moisture content constructed in this study can provide theoretical support for the prediction of small-scale fuel moisture content in Chongli District.