Land surface temperature (LST) is an essential climate variable (ECV) which can be estimated from appropriate measurements of the surface thermal infrared (TIR) radiance. LST varies on a very short time scale and closely depends on the illumination and scan angles considered. To fully exploit LST products, a method for reconstructing the temporal profile and the angular dependence at the same time is proposed here. A combined visible-thermal envelope method (VT-KDTC) is built using kernel-driven (KD) and diurnal temperature cycle (DTC) models, referring to the surface structure and thermal factors, respectively. To demonstrate the reliability of the approach, TIR data from the geostationary satellite Himawari 8 are combined with visible and near-infrared (VNIR) data from the polar orbit satellite Sentinel-3A/3B. In addition to satellite observations, a synthetic dataset from the Soil Canopy Observation, Photochemistry and Energy Fluxes (SCOPE) model is also generated. Considering an anisotropy model in addition to the DTC model leads to a method displaying a better ability to simulate LSTs with a root mean squared error (RMSE) of 0.48 K against the original satellite results, compared to only the DTC model up to 1.44 K. By utilizing the field measurements as a reference, the reconstructed results are improved with a total bias of 0.72K and an RMSE of 2.58 K. Compared to the original results without correction, approximately 41% and 10% decreases are obtained in bias and RMSE, respectively. Our proposed method can also achieve LST downscaling supported by the higher spatial resolution of VNIR data when the temperature difference is assumed to be homogeneous within the coarse pixels. Thus, a simple achievable solution can be used for temperature reconstruction to enhance the quality of the LST product.

中文翻译：

---

地表温度角度和时间重建的综合方法


陆地表面温度 (LST) 是一个重要的气候变量 (ECV)，可以通过适当测量表面热红外 (TIR) 辐射率来估算。 LST 在非常短的时间范围内变化，并且很大程度上取决于所考虑的照明和扫描角度。为了充分利用 LST 产品，本文提出了一种同时重建时间剖面和角度依赖性的方法。使用核驱动（KD）和昼夜温度循环（DTC）模型建立了组合可见热包络法（VT-KDTC），分别参考表面结构和热因素。为了证明该方法的可靠性，将来自地球静止卫星 Himawari 8 的 TIR 数据与来自极轨卫星 Sentinel-3A/3B 的可见光和近红外 (VNIR) 数据结合起来。除了卫星观测之外，还生成了土壤冠层观测、光化学和能量通量 (SCOPE) 模型的综合数据集。除了 DTC 模型之外还考虑各向异性模型，该方法显示出更好的模拟 LST 的能力，与原始卫星结果相比，均方根误差 (RMSE) 为 0.48 K，而仅 DTC 模型高达 1.44 K。以现场测量为参考，重建结果得到改善，总偏​​差为0.72K，RMSE为2.58 K。与未经校正的原始结果相比，偏差和RMSE分别降低了约41%和10%，分别。当假设粗像素内的温差是均匀的时，我们提出的方法还可以实现由 VNIR 数据的更高空间分辨率支持的 LST 缩小。 因此，可以使用简单可实现的解决方案进行温度重建，以提高 LST 产品的质量。