基于机载高光谱影像的植被冠层叶绿素反演

Inversion of vegetation canopy's chlorophyll content based on airborne hyperspectral image

利用黑龙江省伊春市带领区凉水国家级自然保护区机载高光谱数据,提取了红边面积、三角形植被指数、归一化植被指数等15个光谱参数,结合坡度、坡向、海拔、郁闭度和植被总盖度5个地理参数,并利用叶绿素计SPAD-502对研究区植被冠层叶绿素相对含量进行同步测量,分析了叶片光谱反射率、反射率的一阶导数及其他变形分别与SPAD值的相关性,采用基于核变换的偏最小二乘原理建立了叶绿素相对含量的估测模型,用该模型对研究区植被冠层叶绿素相对含量进行定量估算.结果表明:当分段数为3、提取的主成分数为10时,所建模型的效果较好,模型决定系数达到0.855,平均绝对百分误差为9.6%,预测精度为89.7%.

By using the airborne hyperspectral remote sensing data of Liangshui National Nature Reserve in Yiehun of Heilongjiang Province, Northeast China, 15 spectral parameters including red edge area, triangular vegetation index, and normalized difference vegetation index, etc. were extracted, and in combining with 5 geographical parameters including slope, aspect, elevation, canopy density and total vegetation coverage, and by using SPAD-502, the vegetation canopy＇ s relative chlorophyll content in the reserve were measured, with the correlations of the leaf spectral reflectivity, its first-order derivative and other deformations with the SPAD value analyzed. A prediction model for relative chlorophyll content was established by adopting the kernel-based partial least- squares regression, and a quantitative estimation of the vegetation canopy＇ s relative chlorophyll content in the study area was carried out with the established model. The results showed that the model performed best when the sections were three and the principle components were ten. The coefficient of determination of the model was R^2 = 0.855, the mean absolute percent error was 9.6%, and the prediction precision was 89.7%.