Rivers are among the most dynamic components in Earth's terrestrial water cycle and provide critical ecosystem services. Yet, the spatiotemporal variability of river surface extents remains largely unquantified at the global scale. Satellite remote sensing provides a promising alternative to in-situ observations, which can enable a more comprehensive survey and systematic analysis of global rivers at fine spatial resolutions. The study examines the spatiotemporal variability of river surface extent globally and its natural driving factors, by combining the use of Landsat-based Global Surface Water (GSW) and Global River Widths from Landsat (GRWL) databases. In addition to examining the long-term mean river surface extent in various climate zones, we perform statistical analyses to correlate monthly times series of fractional river extent with the terrestrial water storage (TWS) components obtained from the Gravity Recovery and Climate Experiment (GRACE) satellite observation and the Global Land Data Assimilation System (GLDAS) model simulations. Results show that the spatiotemporal variability of water presence in rivers can be explained well via differentiating climate zones. The analysis also shows that 52.7% of the global maximum river extent is covered by water less than half of time. Changes of fractional river extent are found to be highly correlated with groundwater storage in low- and mid-latitudes, whereas snow melting dominates the river dynamics in high latitudes. By examining the extremes of fractional river extent, we found that the abrupt changes of fractional river extent are well linked to precipitation anomalies in the equatorial, arid, and warm temperate areas. This study offers an innovative perspective to study spatiotemporal dynamics of rivers by combining optical remote sensing (Landsat), gravimetry observations (GRACE), and land surface simulations; and it highlights the significant role of low-flow-generating processes (snow melting, infiltration, and recharge-discharge) in controlling river dynamics in certain regions, which warrants future investigation.

河流是地球陆地水循环中最具活力的组成部分之一，并提供关键的生态系统服务。然而，在全球范围内，河流表面范围的时空变化在很大程度上仍未量化。卫星遥感为原位观测提供了一种很有前景的替代方法，可以在精细的空间分辨率下对全球河流进行更全面的调查和系统分析。该研究结合使用基于 Landsat 的全球地表水 (GSW) 和来自 Landsat (GRWL) 数据库的全球河流宽度，研究了全球河流表面范围的时空变异性及其自然驱动因素。除了检查不同气候区的长期平均河流表面范围外，我们进行统计分析，将部分河流范围的月度时间序列与从重力恢复和气候实验 (GRACE) 卫星观测和全球土地数据同化系统 (GLDAS) 模型模拟中获得的陆地蓄水 (TWS) 分量相关联。结果表明，通过区分气候带可以很好地解释河流中水存在的时空变异性。分析还表明，全球最大河流范围的 52.7% 被水覆盖的时间不到一半。发现部分河流范围的变化与低纬度和中纬度的地下水储存高度相关，而融雪在高纬度的河流动态中占主导地位。通过检查部分河流范围的极端情况，我们发现，部分河流范围的突变与赤道、干旱和暖温带地区的降水异常密切相关。本研究通过结合光学遥感 (Landsat)、重力观测 (GRACE) 和地表模拟，为研究河流时空动态提供了一个创新视角；它突出了低流量产生过程（融雪、入渗和补给-排放）在控制某些地区河流动态方面的重要作用，值得进一步研究。