以内蒙古大兴安岭1987年和2003年火烧迹地为研究区,以白桦叶片为研究对象,筛选与白桦叶片氮含量相关性较强的光谱波段,利用多元逐步回归和非线性回归方法构建白桦叶片氮含量的较优预估模型。结果表明:1)1987年和2003年火烧迹地的白桦叶片氮含量分别为25.22和17.23g/kg,2个样区叶片光谱曲线走势较一致,其差异仅表现在“绿峰”与“红边”处;2)根据叶片氮含量与实测光谱及其一阶导数的相关系数,筛选出与氮含量高度相关的波段数分别为67个(1987年)和40个(2003年);3)利用相关性较高的波段作为自变量构建白桦叶片氮含量的多元逐步回归模型和非线性回归模型,多元逐步回归模型较非线性模型的预测效果更佳。氮含量变化会影响植物的光谱敏感度,氮含量越多,绿光波段和近红外波段处的光谱反射率越低;多元逐步回归模型可以作为大兴安岭火烧迹地白桦叶片氮含量的预测模型。

Taking the burned areas in Daxing'anling of Inner Mongolia in 1987 and 2003 as the study area and Betula Platyphylla leaves as the research object,the spectral bands with strong correlation with the nitrogen content were selected,and the optimal prediction model of nitrogen content was constructed by multiple stepwise regression and nonlinear regression methods. The results showed that: 1) the nitrogen content of Betula platyphylla leaves in burned areas of 1987 and 2003 were 25.22 and 17.23 g/kg,respectively. The trend of leaf spectral curves in two areas was consistent,and the difference was only reflected in the "green peak" and "red edge". 2) According to the correlation coefficients between leaf nitrogen content and measured spectrum and its first derivative,67 bands (1987) and 40 bands (2003) were selected. 3) The multiple stepwise regression model and nonlinear regression model of nitrogen content of Betula platyphylla leaves were established by using the high correlation bands as independent variables. The prediction effect of multiple stepwise regression model was better than that of nonlinear regression model. The change of nitrogen content will affect the spectral sensitivity of plants. The more nitrogen content,the lower the spectral reflectance at green band and near infrared band was;The multiple stepwise regression model can be used to predict the nitrogen content of Betula platyphylla leaves in the burned area of Daxing'anling.