Land surface temperature (LST) is listed as an essential climate variable (ECV) and supports quantitative estimates of the energy budget while serving as a proxy for measuring the effects of climate change and extreme events. Forested areas are considered a major land unit impacted by temperature rise; therefore, thorough monitoring is mandatory. An accuracy assessment of the LST of forests must consider their directional anisotropy (DA). This latter can be well depicted by thermal infrared (TIR) radiative transfer models, but the problem is complex for forests because many of the shaded areas generate multiscale gradients of temperature. In this paper, we adapted a mature and widely used visible and near-infrared (VNIR) radiative transfer model called forest reflectance and transmittance (FRT) to enhance the characterization of the DA of forest temperature. In the FRT model, the vertical heterogeneity of the forest is quantified by using the discrete elements of multilayer scene components (i.e., the tree crown, trunk, understory vegetation, and soil), thus inferring vertical thermal gradients. The Planck function and spectral-invariant theory are considered to assess the thermal emissions of the scene components and their multiple scattering processes. The FRT model is validated using directional forest brightness temperatures (BT) measured from an unmanned aerial vehicle (UAV) and simulated by using the three-dimensional ray-tracing LESS (large-scale remote sensing data and image simulation framework over heterogeneous 3D scenes) model. The results show that FRT behaves reliably since the root mean square error (RMSE) is lower than 1.0 °C for UAV measurements obtained at 09:20 and 13:10 and with coefficients of determination (R²) larger than 0.74 and 0.56, respectively; these results are better than the simulated results by existing models. Moreover, the comparison with ray-tracing simulations was also deemed satisfactory. According to the analysis, large variations in BT DAs may appear for different forests and seasonal changes staged by structural and thermal stratification, thus indicating the necessity of using the FRT model for complex and dynamic forest canopies.

地表温度 (LST) 被列为基本气候变量 (ECV)，支持能源预算的定量估计，同时作为衡量气候变化和极端事件影响的代理。林区被认为是受气温上升影响的主要土地单位；因此，必须进行彻底的监测。森林 LST 的准确性评估必须考虑它们的方向各向异性 (DA)。后者可以通过热红外 (TIR) 辐射传输模型很好地描述，但对于森林来说，这个问题很复杂，因为许多阴影区域会产生多尺度的温度梯度。在本文中，我们采用了一种成熟且广泛使用的可见光和近红外 (VNIR) 辐射传输模型，称为森林反射率和透射率 (FRT)，以增强森林温度 DA 的表征。在 FRT 模型中，森林的垂直异质性是通过使用多层场景组件（即树冠、树干、林下植被和土壤）的离散元素来量化的，从而推断出垂直热梯度。普朗克函数和光谱不变理论被认为是评估场景组件的热辐射及其多次散射过程。FRT 模型使用从无人机 (UAV) 测量的定向森林亮度温度 (BT) 进行验证，并使用三维光线追踪 LESS（异构 3D 场景上的大规模遥感数据和图像模拟框架）进行模拟。模型。结果表明，FRT 表现可靠，因为对于在 09:20 和 13:10 获得的 UAV 测量值以及确定系数（R ² ) 分别大于 0.74 和 0.56；这些结果优于现有模型的模拟结果。此外，与光线追踪模拟的比较也被认为是令人满意的。根据分析，不同森林的 BT DAs 可能会出现较大的变化以及结构和热分层阶段的季节性变化，从而表明使用 FRT 模型来处理复杂动态的森林冠层的必要性。