LAUSR

Abstract Rivers are essential to the Earth's ecosystem, but the current understanding of river width variability is limited, owing to the sparse distribution of gauging stations. Remote sensing data enable the surveying and analysis of river geomorphology by providing multi-temporal Earth observation data from satellites at fine spatial and temporal resolutions. We proposed an optimized RivWidth method to automatically calculate width for all channels in a water map and parallelized it to produce the Multi-temporal China River Width (MCRW) dataset, which is the first 30-m multi-temporal river width dataset for China during 1990–2015, including estimates under both seasonal fluctuations and dynamic inundation frequencies. The MCRW dataset is made up of 1.3 × 108 seasonal estimates of river width, and covers 1.4 × 105 km of rivers in China. We validated the MCRW dataset against in-situ measurements. The MCRW estimates at maximum water extent showed a satisfactory accuracy of 15.0% and 15.2% for the mean absolute percentage error (MAPE) and the relative root-mean-square error (RRMSE), respectively. The MCRW dataset was further compared with the current state-of-the-art global product, the Global River Widths from Landsat (GRWL) dataset, which demonstrated the superiority of the MCRW in describing the basins of China. Our analysis indicated that the mean river widths of China in both summer and winter have increased over the past 25 years, and river width of the Yangtze River mainstream in the lower drainage region has shown a downward trend while the its middle reaches and tributaries (upstream of the Three Gorges Dam) have shown an upward trend. We also developed a locally adaptive search method to quantify seasonal (summer and winter) river width variability. The results revealed that most of the rivers were wider in summer during the study period, and mainstream of Yangtze River in middle/lower region exhibited less seasonal variability than its tributaries. Larger widths were observed in the middle reaches of the Yellow River and the upper reaches of the Black River in winter due to ice-jam floods. Overall, the generated MCRW dataset has the potential to serve as a fundamental resource in Earth system science, and could provide valuable support to surface water resource and riverine management.