AbstractSoil erosion is a big concern in bare soils from burnt areas and agricultural lands that lack a vegetation cover. In those unprotected soils, intense rain episodes, typical in Mediterranean climate, cause severe soil erosion processes that have been well studied previously using a number of procedures, such as the geomorphic change detection (GCD) method. This method uses digital elevation models (DEMs) of the soil surface and determines the morphological changes in terms of both erosion and deposition by DEMs of difference (DoDs). However, some types of soil erosion, such as diffuse and sheet erosion, may have a small magnitude, at a millimetre scale, and their determination requires methods adapted to that scale. In this paper, we analyse the suitability of the GCD method to account for small magnitude soil erosion. We present a laboratory procedure and setup to represent and quantify sediment budget in small experimental soil plots through differences of DEMs obtained from images using photogrammetric structure from motion. This study explores several key aspects of the technique: establishment of a common reference system for DEMs; determination of errors in the generation of DEMs; selection of appropriate criteria to obtain reliable changes in DoDs; error propagation using Monte Carlo simulation; and validation of the procedure by comparing the results with actual sediments collected during the experiment. Results showed an overestimation of 13% in accumulated soil loss and confirmed that GCD approach with structure from motion‐based DoDs is a suitable method to quantify small‐magnitude erosion events.