LAUSR

Assumptions about fluvial processes and process–form relations are made in general models and in many site‐specific applications. Many standard assumptions about reach‐scale flow resistance, bed‐material entrainment thresholds and transport rates, and downstream hydraulic geometry involve one or other of two types of scale invariance: a parameter (e.g. critical Shields number) has the same value in all rivers, or doubling one variable causes a fixed proportional change in another variable in all circumstances (e.g. power‐law hydraulic geometry). However, rivers vary greatly in size, gradient, and bed material, and many geomorphologists regard particular types of river as distinctive. This review examines the tension between universal scaling assumptions and perceived distinctions between different types of river. It identifies limits to scale invariance and departures from simple scaling, and illustrates them using large data sets spanning a wide range of conditions. Scaling considerations and data analysis support the commonly made distinction between coarse‐bed and fine‐bed reaches, whose different transport regimes can be traced to the different settling‐velocity scalings for coarse and fine grains. They also help identify two end‐member sub‐types: steep shallow coarse‐bed ‘torrents’ with distinctive flow‐resistance scaling and increased entrainment threshold, and very large, low‐gradient ‘mega rivers’ with predominantly suspended load, subdued secondary circulation, and extensive backwater conditions. © 2020 The Authors. Earth Surface Processes and Landforms published by John Wiley & Sons Ltd