In this experimental work, flow force in compound channel in presence of non-submerged vegetation was obtained via direct measurement method. The direct measurement method was used to address the inefficiency of the conventional energy-based methods for measuring the force within the body of highly rough flows. On this basis, the study was performed using a specially designed flume called knife-edge flume with a length of 14 m, a width of 1.07 m, and a height of 1.05 m at a stream bed grade of 10-3. The flow force measurements were performed directly using a load cell in absence of the common errors that were otherwise associated with velocity measurement by inserting an external body into the flow. In this study, we examined non-submerged rigid vegetation layers with five different diameters (i.e., 20, 25, 30, 40, and 50 mm), three values of crossline spacing (i.e., 6, 8, and 10 cm), five values of crossline spacing (i.e., 8, 10, 12, 15, and 20 cm), and two different arrangements (i.e., ordered and non-ordered or cruciform) at three different positions (i.e., in the floodplain, in the main channel, and simultaneously in the floodplain and the main channel) by performing a total of 451 tests. The flow force exhibited the highest sensitivity to the increase in vegetation volume under the non-ordered arrangement. In this respect, the improvement in the flow force with increasing the percent volume of vegetation was much larger when the vegetation was arranged in a non-ordered rather than ordered fashion, with the effect of the vegetation been even more pronounced in the floodplain rather than the main channel. In addition, at a certain percent change of the flow force, the trough-vegetation flow velocity was higher under the ordered arrangement rather than the non-ordered arrangement, reflecting the more difficulty encountered by the flow as it passed through the non-ordered vegetation. Based on the analysis results, it was found that the flow force would change by 060% when the vegetation had grown in the floodplain with λ* values in the range of 0–1. In cases where the vegetation had grown in the main channel and simultaneously in the main channel and the floodplain, the flow force variations increased to 10–70% and 20–80% as the λ* changed from 0 to 10, respectively. Analyzing the results of the experiments and the vegetation-modified Reynolds number, a relationship was further presented for calculating the flow force in presence of non-submerged rigid vegetation in a compound channel.