The development of bed load calculation methods directly depends on the reliability of the measurement data. The most reliable measurement data remains the data obtained by the volumetric method when observing the filling of reservoirs, borrows, ditches etc. Nevertheless these data are the rarest. In this paper on the base of the data obtained when observing the process of filling of a ditch across the Amur River a comparison of a number of bed load calculation methods is performed. The observations were carried out with a multi-beam echo-sounder during summer floods of 2018, from 21st of July to 22nd of August. Over this time 5 surveys were performed, that allows to have 4 calculation periods for determining bed load yield. The total number of the measurements at different calculation verticals is 108. These data are used for verification of 80 bed load formulas. Four methodological approaches are considered: bed form approach, critical velocity approach, critical water discharge approach and regression approach. The bed form approach has shown the greatest accuracy: 17 formulas out of 26 gave the error less than 60%. For the other 56 methods which were considered only 5 formulas showed the error less than 60%, all of them correspond to the critical velocity approach.