Abstract: Palaeoflood hydrology is a newly developed frontal subject in global change science. There several models could be used to calculate palaeoflood discharge in the bedrock channel according to the palaeoflood slackwater deposits （SWD），but the channel roughness is a complex sensitivity parameter in flood discharge calculation. Taking 4 period palaeofloods recorded in the Shangjiahe platform frontier in the Yun County bedrock reach of the upper reaches of the Hanjiang River as an example，the influences of the channel roughness coefficient were studied by using the slope-area method and HEC-RAS model to calculate the palaeoflood discharges，respectively. The results showed that when other hydrologic parameters didn’t change，and if the channel comprehensive roughness coefficient only changed about ±15%，the errors of palaeoflood discharges were -13.04%～17.65% with the slope-area method，and -4.32%～4.57% with HEC-RAS model，respectively. And if the channel comprehensive roughness coefficient only changed about ±25%，the errors of palaeoflood discharges were-20%～33.33% using the slope-area method and -7.79%～9.18% using HEC-RAS model to reconstruct palaeoflood discharges，respectively. This indicated that HEC-RAS model could effectively reduce the influence of roughness coefficient to reconstructe palaeoflood discharges，and the palaeoflood discharges calculation was more reliable. Moreover，if the roughness coefficient of the main channel，left bank and right bank changed ±25% respectively，the discharge change of the main channel roughness coefficient has the greatest influence on the results of discharge calculation. This also indicated that it was very important to select accurately the main channel roughness coefficient to calculate flood discharge. The results provided a theoretical basis for the hydrological engineering calculation and flood control and disaster mitigation.

Key words: roughness coefficient, slope-area method, HEC-RAS Model, palaeoflood discharge calculation