­
Figure 3. a) Global riverbank erosion rates averaged at Pfafstetter level 6 basins. b) Amazon basin inset showing Pfafstetter level 6 averages. c) reach averages using the approximately 10 km reaches in SWORD. d) The finest scale of REAL data shown for a section of the Purus River, node data spaced approximately every 200m.
Beyond the broad, global scale spatial patterns there is a lot of potential analysis with regards to inter- and intra-basin relationships. We show scatterplots of river width versus the observed bank erosion rate for the 16 basins with the most valid observations to investigate the first order scaling relationships (Figure 4). Generally, the relationships show a spectrum from highly linear (e.g. Ganges-Brahmputra-Meghna) to weak relationships and clustered values near the low end of width and erosion rate (e.g. Yangtze and Yenisei). In comparison to the global relationship proposed by Hooke (1982) shown in dark red, some basins approximate this relationship well (e.g. Amazon and Paraná), while several notable river basins are biased low (e.g. Congo, Ob’, and Yukon).