热红外核驱动模型在垄行植被热辐射方向性特征拟合中的精度分析

Ridge planting is a common planting structure for orchards and crops. Its canopy has stronger thermal radiation anisotropy than continuous and discrete vegetation canopies. Conducting thermal radiation directionality correction for ridge-row vegetation canopies is of great significance for applications such as drought monitoring. The thermal infrared kernel-driven model is the basis for thermal radiation directionality correction, but the fitting accuracy of existing kernel-driven models in ridge-row vegetation canopies has not been discussed. Based on the measured directional characteristics of two groups of vineyard canopies with different orientations obtained in Bordeaux, France, this paper comparatively analyzes the fitting accuracy of 8 existing models (including the Ross - Li and LSF - Li models belonging to the extended framework of visible and near-infrared kernel-driven modeling, and the Vinnikov, RL, Vinnikov - RL, Vinnikov - Chen, LSF - RL, LSF - Chen models belonging to the general framework of thermal infrared kernel-driven modeling). The results show that none of the existing models can fit the row effect characteristics of ridge-row vegetation, the overall accuracy is low (RMSE is as high as 2 K, R² is lower than 0.7), and the fitting abilities of the various models have little difference. None of the existing thermal infrared kernel-driven models are applicable to ridge-row vegetation canopies, and it is necessary to develop new kernels that can characterize ridge-row features under the general framework of kernel-driven modeling.

垄行种植是果园及农作物的常见种植结构,其冠层比连续和离散植被冠层具有更强的热辐射各向异性,开展垄行植被冠层的热辐射方向性校正对干旱监测等应用具有重要意义。热红外核驱动模型是热辐射方向性校正的基础,但现有核驱动模型在垄行植被冠层中的拟合精度未被讨论过。本文以在法国波尔多获取的两组不同朝向葡萄园冠层航空实测方向性特征为基础,对比分析了现有8个模型的拟合精度(含隶属可见光近红外核驱动建模扩展框架的Ross-Li、LSF-Li模型及隶属热红外核驱动建模通用框架的Vinnikov、RL、Vinnikov-RL、Vinnikov-Chen、LSF-RL、LSF-Chen模型),结果表明现有模型均无法拟合出垄行植被的行效应特征,精度整体较低(RMSE高达2 K,R~2低于0.7)且各个模型之间的拟合能力差异小,现有热红外核驱动模型均不适用于垄行植被冠层,有待在核驱动建模通用框架下发展可刻画垄行特征的新核。