哈尔滨城市林业示范基地5种典型人工林地表死可燃物含水率的预测

Prediction on Moisture Content of Surface Dead Fuels in Five Typical Artificial Forests in Harbin Urban Forestry Demonstration Base

在哈尔滨城市林业示范基地内,以白桦、黑皮油松、蒙古栎、水曲柳、兴安落叶松等5种典型人工林为研究对象,将地表死可燃物含水率作为独立因子,与林内相关性显著的气象因子进行回归分析,采用气象要素回归法和随机森林法建立含水率预测模型,并对两种预测方法的模型误差进行分析。结果表明:5种典型人工林地表死可燃物含水率均与温度、相对湿度、前12 h降水量极显著相关,其中蒙古栎林与太阳辐射紫外光有显著关系,兴安落叶松林与风速有显著关系;气象要素回归法的平均绝对误差和均方根误差范围分别是12.67%~21.36%和18.07%~49.23%,随机森林法的平均绝对误差和均方根误差范围分别是8.35%~9.50%和15.71%~19.48%。两种方法建立气象要素的地表死可燃物含水率预测模型,对于白桦、蒙古栎和水曲柳预测精度较差,黑皮油松和兴安落叶松预测精度较高。对比气象要素回归法,随机森林法预测准确性更高,不同林分之间误差更小。

In Harbin Urban Forestry Demonstration Base,five typical plantations of Betula platyphylla,Pinus tabuliformis var.mukdensis,Quercus mongolica,Fraxinus mandshurica and Larix gmelinii were studied,taking the moisture content of surface dead fuel as an independent factor,the regression analysis was carried out with the meteorological factors with significant correlation in the forest,the moisture content prediction model was established by using meteorological factor regression method and random forest method,and the model errors of the two prediction methods were analyzed.The moisture content of surface dead fuels in five typical artificial forests was significantly related to temperature,relative humidity and rainfall in the first 12 h.Q.mongolica forest was significantly related to solar ultraviolet radiation,and L.gmelinii forest was significantly related to wind speed.The MAE and RMSE ranges of meteorological element regression method were 12.67%-21.36% and 18.07%-49.23%,respectively,and the MAE and RMSE ranges of the random forest method were 8.35%-9.50% and 15.71%-19.48%,respectively.In the prediction models of surface dead fuels moisture content of meteorological elements established by the two methods,the prediction accuracy of B.platyphylla,Q.mongolica and F.mandshurica is poor,while that of P.tabuliformis var.mukdensis and L.gmelinii is high.Compared with the regression method of meteorological elements,the prediction accuracy of random forest method is higher,and the error between different stands is smaller.