In this study, an assessment of similarity relationships across runoff measurements at various spatial scales was carried in a typical Sahelian landscape, under semi-arid climate in northern Burkina Faso (west African Sahel). The scales of observations considered are the plot scale (1 m2, 50 m2, 150 m2) and the sub-catchment scale (6.1 ha, 33.8 ha), with various soil surface conditions and slopes. These scales were monitored during six years (2010–2015) under natural rainfall. Based on monitoring data, a runoff potential index I1 and an effective runoff index I2 are developed, encapsulating the physical characteristics of the observation scale into dimensionless forms, hence reducing their sensitivity to the scale effects. These indices help in assessing hydrological similarity across measurement scales. Results showed that runoff decreases with the increase of the observation scale size. Besides, dimensionless indices vary following a logarithmic decay relationship with the scale size (R2 > 0.98). Such results highlight the non-linear nature of runoff processes and also promotes similarity analysis as a means to model hydrological processes through soil surface conditions and landscape descriptors.