考虑到植被可见光-近红外的光谱吸收特征与光合有效辐射吸收率(fraction of absorbed photosynthetically active radiation, FAPAR)有很好的关联, 综合“高光谱曲线特征吸收峰自动识别法”与“光谱吸收特征参量化法”, 提取对FAPAR敏感的高光谱吸收特征参数, 借鉴可见光-近红外植被指数的数学形式, 尝试用优化组合后的可见光-近红外光谱吸收特征参数替代光谱反射率, 构建新型植被指数估算植被FAPAR, 并利用2014年和2015年内蒙古自治区中部与东部地区天然草地典型群落冠层实测光谱数据进行FAPAR估算建模与验证。 结果表明: 新型植被指数“SAI-VI”不仅有效提高了单个光谱吸收特征参数在高、 低覆盖区域估算FAPAR的精度, 而且相比五种与FAPAR有较好相关性的具有不同作用类型的可见光-近红外植被指数, 其与FAPAR值的相关性更高（存在最大相关系数=0.801）, 以其为变量的指数模型预测FAPAR精度更高且稳定性较好（建模与检验的判定系数均最高且超过0.75, 标准误差与平均误差系数也相应最小）。 研究表明: 融入可见光-近红外高光谱吸收特征的新型植被指数“SAI-VI”, 强化了可见光波段与近红外波段光谱吸收特征的差别, 相较单一光谱吸收特征参数, 在降低土壤背景影响的同时增强了对FAPAR变化的敏感度。 同时, “SAI-VI”有效综合了对植被FAPAR敏感的光谱吸收特征信息, 相较原始光谱反射率, 能表达植被光合有效辐射吸收特征的更多细节信息, 可作为植被冠层FAPAR反演的新参数, 一定程度上弥补当前植被指数法估算FAPAR的不足。

Considering the close relationship between spectral absorption features of visible-near infrared and “Fraction of Absorbed Photosynthetically Active Radiation(FAPAR)”, the “automatic recognition method of hyperspectral curve’s characteristic absorption peak” and “quantization method of spectral absorption characteristic parameters” were used to extract the hyperspectral absorption characteristic parameters which are sensitive to FAPAR. Referring to mathematical form of vegetation index, visible-near infrared spectral absorption characteristic parameters were used to replace spectral reflectance and create a new vegetation index to estimate FAPAR of vegetation. The data from 2014 and 2015 on typical natural grassland community canopy in the middle and eastern Inner Mongolia was chosen to build and verify the model of estimating FAPAR. The results showed that new vegetation index “SAI-VI” effectively raised the FAPAR estimating accuracy in the middle and low vegetation coverage areas. Compared with other seven different types of visible-near infrared vegetation index, it has a higher correlation with the value of FAPAR(the largest correlation coefficient is 0.801). The FAPAR prediction index model which takes “SAI-VI” as variable has higher precision and better stability(the determination coefficients of modeling and testing are the highest and both are above 0.75, the “Root Mean Square Error (RMSE)” and “Average Error Coefficient (MEC)” are the minimum). The research also showed that the new vegetation index “SAI-VI” infusing visible-infrared spectral absorption feature highlights the difference between visible spectral and near infrared spectral absorption characteristic parameters. While comparing with single spectral absorption characteristic parameter, “SAI-VI” can depress the influence of soil and enhance the sensitivity to the changes of FAPAR. “SAI-VI” also included the information of hyperspectral absorption characteristic parameters which are sensitive to FAPAR and expressed more detailed information of FAPAR while comparing with original spectral reflectance. “SAI-VI” can be used as a new parameter in inversion of vegetation canopy FAPAR, to some extent it could remedy defect of vegetation index method in estimating FAPAR.