Abstract: The broad valley reach of the Yarlung Zangbo River presents unstable multiple channels and ephemeral bars formed by intense bed-load transport and active morphodynamic processes, exhibiting notable braided river characteristics. This study uses multi-source data comprehensively－including field observations and satellite remote sensing, adopts an artificial neural network model to classify and identify the satellite images－and determines statistically the geometric parameters of the river and its bars. We examine the time evolution of characteristic geomorphic factors as a response to flow conditions in well-braided reaches, and work out a quantitative relationship for the response. The results indicate braided river geomorphological parameters are highly sensitive to flow variations and display pronounced seasonal variations. The number of bars increases with the rising discharge, while their total area decreases; The bar braid index (BIb) is positively correlated with the flow rate, and the reach braid index (BIr) shows a non-monotonic change－decrease in flood peak period, rise in flood retreat period, then decrease again in dry season. The results deepen our understanding of complex river morphological evolution, and help develop and utilize plateau rivers.

Key words: braided river, Yarlung Zangbo River, morphological evolution, remote sensing, braiding index