Human activities have increased rapidly in the last five decades and introduced changes to the flow-sediment regime, particularly the sediment supply, in the downstream reach, inevitably influencing the evolutionary processes of the gravel-sand transition (GST). The Yangtze River, the third-largest river in the world, appears to represent a field example of a GST with strong human interference. Since 1975, the sediment transport in the Yangtze River has been frequently disturbed by human activities, whereas there is no systematic study on the reach-scale morphodynamics of the gravel-sand transition.

We utilized a combination of long-term (1975–2018) field observations (topographic data and bed material information) and a 2D hydrodynamic model from Yangtze GST, to study channel morphodynamic processes and their controls with respect to anthropogenic modification. The results showed that (1) Over the past 50 years, except for the period 1991~2002 when the channel was in quasi-equilibrium, the Yangtze GST has experienced continuous adjustments and intensive erosion. The channel gradient became steeper and the width became larger, but morphology variability continued to decrease. Despite these adjustments, the spatial pattern of channel morphology remained stable; the channel gradient and the bankfull width of Zhicheng-Chenjiawan Region exhibited a high and protruding level, with 2~3 times that of the upper and lower regions; (2) As the bed surface coarsened, the location of the Yangtze GST has undergone a migration process of “stable→downwards→stable”. The downstream migration happened from 2003 to 2010, with the upper and lower boundaries of the gravel-sand transition migrating downstream by 48.9 km and 10 km, respectively. The grain size distributions within it transformed from mainly bimodal gravel-sand mixtures to predominantly spatial segregation of grain sizes with alternating sand and gravel patches; (3) Due to the morphological control, the Yangtze GST exhibited a poorly understood phenomenon in which flow competence was much greater at low flow than at flood peak somewhere, with the water surface slope ranged from 0.9 ‱ to 0.3‱ at flood peak but reached a high level of up to 3.5‱ at low flow. As the base level at Shashi station decreased, the maximum water surface slope at low flow became greater. Our results provide not only a theoretical basis for channel adjustment prediction but also a powerful reference for channel regulation of the Middle Yangtze River.