Avulsion of the Brahmaputra in Bangladesh during the 18th–19th century: A review based on cartographic and literary evidence

Avulsions involve switching of fluvial discharge through a new channel from an existing course and occur over varied timeframes ranging from hours to decades. With a peak discharge above 100,000 m³/s, the Brahmaputra (2900 km) is one of the largest rivers in the world. Its lower course through the Ganga–Brahmaputra–Meghna (GBM) Delta of Bangladesh was first mapped in 1764–73 as a broad eastward-swinging curvature skirting the subsiding Sylhet Wetlands and falling into the Meghna River. Beginning from the late eighteenth century, the river started to avulse into a linear, more direct, southward passage into the Ganga called the Jamuna and decreased its length by 46 km (14.2%) from its former course to the Bay of Bengal. Examination of the established timeline and procedure of the avulsion event using maps and images from 1764–73, 1830–34, 1854–60, 1907–24, and 2014, in addition to evidence from contemporary literature and tree-ring reconstructed long-term discharge data, connoted certain discrepancies. Our re-evaluation indicated that the Brahmaputra shifted westward and was captured by a parallel distributary of the neighbouring Tista system initiating the switch. This was probably aided by exceptionally high monsoonal discharge between 1786 and 1790 as well as the 1787 break-in of the Tista into the Brahmaputra. Because of its high width:depth ratio and braided planform that prevented fast escape of the discharge, abandonment of the eastern channel—the ‘Old’ Brahmaputra—progressed gradually; well into the next century with the new channel taking over almost completely by 1855–60. Comparable switchings of the Brahmaputra occurred six times in the last 11 kyr and significantly influenced development of the GBM Delta. The century-scale timeline of the last avulsion event described here refines its current understanding by proposing river capture as the mechanism of avulsion, possibly triggered by clustering of high-flood years.