Measurements at bankfull stage were made at 136 sites to determine whether predictions of flood discharges using channel-geometry characteristics are as good as or better than predictions using basin characteristics. These measurements are used to determine the variables of bankfull width, area, depth, and velocity. These variables are combined with basin characteristics for the sites, and multiple-regression techniques are used to select the best combination of variables to estimate the selected floods.
Generally, discharge estimates obtained from equations using channel characteristics have smaller standard errors than those using basin characteristics, and equations using both basin and channel characteristics have even lower standard errors.