联合相似像元算法与地表能量平衡方法的云下地表温度估算研究

Land surface temperature (LST) is one of the most important parameters in the physical processes of surface energy and water balance from local to global scales. However, Thermal infrared remote sensing data are often contaminated by clouds and resulting in failure of retrieving the real LST, which has seriously limited the applications of LST products. Therefore, much attention has been paid to estimating LST under the clouds, and only few of methods can be used successfully. The similar pixel algorithm can obtain the LST of cloudy-pixels only in the cases of hypothetical clear-sky, while the surface energy balance-based physical models can acquire the real LST. However, the method is heavily depending on the long term in-situ observations, thus limiting its application in ungauged region. Besides, the accuracy needs to be further improved. In this project, MODIS MOD11A1/MYD11A1 data sets are chosen as our objectives. First, the similar pixel algorithm is improved by a new scheme of temporal neighboring reference image, the high accuracy hypothetical clear-sky LST values were filled to cloudy-pixels; Second, a surface energy balance method which parameterized by remote sensing data, is used to correct the filled values. And the hypothetical clear-sky LST is corrected to the real LST in cloud-contaminated areas with net shortwave radiation data. Finally, long term ground LST measurements will be used for validation. This project will provide some essential theoretical and technological solutions to cloudy-pixel LST estimation method in ungauged region. It is also valuable for improving the accuracy of current cloudy-pixel LST estimation.

地表温度是区域和全球尺度陆表能量与水分交换过程的一个关键参数。然而云覆盖导致的遥感观测数据缺失，严重限制了地表温度产品的实际应用。现有的云下地表温度估算方法中，相似像元算法得到的是“理论晴空”下的地表温度，而基于地表能量平衡方法的物理模型虽能得到云下地表的真实温度，但依赖于大量地面观测数据实现模型参数化，且估算精度有待提高。本项目以MODIS地表温度数据为研究对象，首先通过参考影像的改进，来提高相似像元算法的稳定性，得到高精度的云下像元“理论晴空”地表温度；然后基于地表能量平衡方法，利用遥感数据来替代站点观测数据的参数化方案，结合短波净辐射数据定量估算云覆盖对地表温度的影响，以此对“理论晴空”地表温度进行校正，得到真实的云下地表温度；最后采用长时间序列站点观测数据进行验证。通过本研究可构建稳定的、能用于无资料地区的云下地表温度重建模型，有望进一步提高云下地表温度估算精度。

地表温度是区域和全球尺度陆表能量与水分交换过程的一个关键参数。然而云覆盖导致的遥感观测数据缺失，严重限制了地表温度产品的实际应用。现有的云下地表温度估算方法中，相似像元算法得到的是“理论晴空”下的地表温度，而基于地表能量平衡方法的物理模型虽能得到云下地表的真实温度，但依赖于大量地面观测数据实现模型参数化，且估算精度有待提高 。本项目以MODIS地表温度数据为研究对象，首先通过滑动平均的方式改进参考影像的生成方案，并自动选取临近时间的参考影像，来提高相似像元算法的稳定性，得到高精度的云下像元“理论晴空”地表温度；然后基于地表能量平衡方法，利用遥感数据来替代站点观测数据的参数化方案，结合短波净辐射数据定量估算云覆盖对地表温度 的影响，以此对“理论晴空”地表温度进行校正，得到真实的云下地表温度；最后采用长时间序列站点观测数据进行验证。通过本项目构建稳定的、能用于无资料地区的云下地表温度重建模型，进一步提高了云下地表温度估算精度。重建后地表温度精度评价表明，夜间的重建精度相对较高，平均偏差0.57 K，RMSE 1.90 K；日间的平均偏差-0.14 K，RMSE 3.16 K。项目发表SCI论文3篇，EI论文1篇。