Global surface temperature has been setting new record highs in the recent decades, imposing increasing environmental challenges for societies and ecosystems worldwide. Global warming rates of the 20th century have been documented by a number of studies, nevertheless, the warming rates in the most recent decades in the 21st century are of particular interest for understanding the ongoing climate change. Analyzing temperature trends demands data with high spatial resolution and broad geographical coverage to allow for analyzing trends and changes on a regional scale. Land Surface Temperature data from NASA MODIS with global resolution of 0.05° and Skin Temperature data from European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5-Land reanalysis with global resolution of 0.1° fulfill these demands. In this study, we analyze the remote-sensing-based MODIS data to estimate land surface temperature change rates over the period 2001–2020 in global, continental, and pixel-wise scales with statistical significance indicated. The model-based ERA5-Land data are also analyzed in parallel, extending the period of analysis back to 1981. These two independently-sourced datasets, one from satellites above the atmosphere and one from combining surface modeling and observations, are shown to produce highly consistent results. It is revealed that the trends in the shorter period 2001–2020 are spatially conforming to the trends in the longer period 1981–2020 despite the shorter time length. For the period 2001–2020, we show that the global average land surface temperature rate of change was 0.26 °C-0.34 °C per decade, with substantially different warming rates in different regions. The Arctic, Europe, and Russia show statistically significant warming in both datasets. The Arctic, in particular, warmed at a rate 2.5–2.8 times the global average, and data in the 40-year period 1981–2020 suggest that warming is accelerating in almost all the continents or large regions. Most noticeably, the two independent datasets both indicate that Arctic permafrost regions had the world's highest warming rate at the onset of the 21st century, reaching >2 °C per decade in some areas.

近几十年来，全球地表温度不断创下新高，给全球社会和生态系统带来了越来越多的环境挑战。许多研究已经记录了 20 世纪的全球变暖速率，然而，21 世纪最近几十年的变暖速率对于了解正在进行的气候变化具有特别重要的意义。分析温度趋势需要具有高空间分辨率和广泛地理覆盖范围的数据，以便分析区域范围内的趋势和变化。来自 NASA MODIS 的全球分辨率为 0.05° 的地表温度数据和来自欧洲中期天气预报中心 (ECMWF) ERA5-Land 再分析的全球分辨率为 0.1° 的皮肤温度数据满足了这些需求。在这项研究中，我们分析了基于遥感的 MODIS 数据，以估计 2001-2020 年期间全球、大陆和像素尺度的地表温度变化率，并表明具有统计意义。基于模型的 ERA5-Land 数据也进行了并行分析，将分析期延长至 1981 年。这两个独立来源的数据集，一个来自大气层上方的卫星，另一个来自结合了地表建模和观测的数据，显示出产生了高一致的结果。结果表明，尽管时间长度较短，但 2001-2020 年较短时期的趋势在空间上与 1981-2020 年较长时期的趋势一致。在 2001-2020 年期间，我们表明全球平均地表温度变化率为每十年 0.26°C-0.34°C，不同地区的升温速率大不相同。北极、欧洲和俄罗斯在两个数据集中都显示出具有统计学意义的变暖。尤其是北极，变暖的速度是全球平均水平的 2.5-2.8 倍，而 1981-2020 年 40 年期间的数据表明，几乎所有大陆或大片地区的变暖正在加速。最值得注意的是，这两个独立的数据集都表明，北极永久冻土区在 21 世纪初的升温速率是世界上最高的，在某些地区每十年达到 > 2 °C。1981-2020 年 40 年的数据表明，几乎所有大陆或大区域的变暖正在加速。最值得注意的是，这两个独立的数据集都表明，北极永久冻土区在 21 世纪初的升温速率是世界上最高的，在某些地区每十年达到 > 2 °C。1981-2020 年 40 年的数据表明，几乎所有大陆或大区域的变暖正在加速。最值得注意的是，这两个独立的数据集都表明，北极永久冻土区在 21 世纪初的升温速率是世界上最高的，在某些地区每十年达到 > 2 °C。