Gullies are an extreme form of soil erosion that degrade diverse environments trough the siltation of streams and water bodies. Indirectly, gully erosion compromises crop productivity working as a link to watercourse allowing movement of detached topsoil particles from agricultural fields during heavy storm events. Furthermore, studies found reduction of the catchment area when active gullies are present. This complex process involves multiple factors and it demands to be studied consistently in order to locate the areas prone for gully erosion. The determination of gullies areas depends upon topographical, geological, and hydrological characteristics; however its location is mainly controlled by the high capacity of overland flow to cut the channel. We hypothesize that identification of gully in agricultural landscape can be performed from high-resolution elevation data products and unsupervised clustering approaches. In order to examine this hypothesis we have used variables resultant from of LiDAR-based terrain analysis as input of a three clustering techniques.    A k-means, fuzzy k-means, and CLARA clustering algorithms were used to carry out the cluster analysis. The results of the cluster analysis suggested that 8 classes were optimal for group areas in the watershed. Elevation data from one field-scale watershed near Treynor in Pottawattamie County, IA, was used to calibration purpose and terrain analysis using slope, flow accumulation, plan convexity, topographic wetness Index, and stream power index were calculated. The cluster analysis has shown highest concordance with percentage of corrected classified pixels that approach based in medoid (CLARA) has obtained the best agreement of points within gullied area (30.1%). The results of this research might speed up gullies field surveys and also can serve as input in conservation planning framework.