State Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China.
State Key Laboratory of Resources and Environmental Information System, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China; University of Chinese Academy of Sciences, Beijing 100049, China.
Sci Total Environ. 2022 Dec 20;853:158406. doi: 10.1016/j.scitotenv.2022.158406. Epub 2022 Aug 30.
The warming amplification of the Tibetan Plateau (TP) has exerted great impacts on the environment in and around the region. It is necessary to thoroughly investigate the temperature changes over the TP. However, the commonly used station observations, satellite products and reanalysis data for relevant studies suffer from deficiencies of sparse spatial distribution, limited temporal coverage and large uncertainties, respectively. This leaves the current understanding of temperature change on the TP still inadequate. Therefore, we propose a multi-source data fusion method to integrate the advantages of different data. Combining the Moderate-resolution Imaging Spectroradiometer (MODIS) land surface temperature (LST) products and station observations, this new method first obtains short-term satellite-derived surface air temperature (SAT) estimates over grids. The European Center for Medium-range Weather Forecasts ReAnalysis 5 land portion (ERA5-Land) temperature with long time range is then clustered and used as independent variables for the Bayesian Ridge Regression (BRR) model. Based on this BRR model, the temporal span of the estimated gridded satellite SAT is extended, and a fused 1-km monthly mean SAT dataset is generated over the TP from 1961 to 2020. The results indicate that the fused data generated by our proposed method has good accuracy with overall RMSE and MBE of 1.33 °C and 1.03 °C, respectively. Despite the temporal and spatial heterogeneity, the performance of the fused SATs is acceptable across seasons and geographical locations. The dataset also shows a great potential for detecting accurate long-term temperature changes across the TP. This fused SAT data owns the advantages from multiple data sources with high accuracy, good spatial continuity, fine spatial resolution and wide temporal coverage, which confirms that our fusion method can provide a favorable opportunity to explore the warming over the TP.
青藏高原（TP）的变暖放大效应对该地区及周边环境产生了巨大影响。有必要深入研究 TP 的温度变化。然而，相关研究中常用的台站观测、卫星产品和再分析数据分别存在空间分布稀疏、时间覆盖有限和不确定性大等缺陷。这使得目前对 TP 温度变化的理解仍然不足。因此，我们提出了一种多源数据融合方法，以整合不同数据的优势。该方法结合中分辨率成像光谱仪（MODIS）陆地表面温度（LST）产品和台站观测数据，首先利用卫星获取短期网格表面空气温度（SAT）估计值。然后，对长时间范围的欧洲中期天气预报中心再分析 5 陆地部分（ERA5-Land）温度进行聚类，并将其作为贝叶斯岭回归（BRR）模型的独立变量。基于该 BRR 模型，扩展了估计的格网 SAT 的时间跨度，并生成了 1961 年至 2020 年 TP 上的融合 1km 月平均 SAT 数据集。结果表明，我们提出的方法生成的融合数据具有较好的精度，整体 RMSE 和 MBE 分别为 1.33°C 和 1.03°C。尽管存在时空异质性，但融合 SAT 在不同季节和地理位置的表现均可以接受。该数据集还显示出在整个 TP 上检测准确长期温度变化的巨大潜力。该融合 SAT 数据集具有多源数据的优势，包括高精度、良好的空间连续性、精细的空间分辨率和广泛的时间覆盖范围，这证实了我们的融合方法可以为探索 TP 的变暖提供有利机会。
