Gully head development represents a significant geomorphic process in a wide range of environments. Several studies investigated the critical topographic conditions, expressed by local slope gradient (s) and drainage area (A), controlling the development and position of gully heads in various landscapes. This review examines over 39 publications. After critically analysing the reported threshold data and after standardisation of the procedure
to determine the critical topographic conditions for gully head development, i.e., sAb N k or s N kA−b some data
sets were discarded because they were not compatible with the standard presentation of data as reported by the majority of studies. Hence, a detailed analysis was made of 63 reported s–A relationships for overland-flow induced gully-heads extracted from data sets collected in various parts of the world. A first examination of the behaviour of both the exponent b and the threshold coefficient k, which reflects the resistance of the site to gully head development, shows clear effects of land use on the value of k whereas the value of b does not seem to be affected. Further analyses are conducted of the recalculated threshold coefficients k, for two predefined constant values of the exponent b. The lowest k-valueswere observed for cropland followed by values for rangeland, pasture and forest. Effects of climate, rock fragment cover at the soil surface and water storage capacity of the gully catchment on k-valueswere also shown. Themost interesting result is that for a given and constant b-value, the threshold coefficient k can be predicted using soil and vegetation characteristics, based on the NRCS Runoff Curve Number values and on surface rock fragment cover.
Furthermore, the underlying physical processes explaining the value of the exponent b were analysed. Finally, a
physically-basedmodel,well anchored in the established theories, is proposed as a first step to predict gully head
development in various landscapes and under changing environmental conditions. The results of this review clearly show that better and more reliable models can be built, including effects of land use, climate changes
and natural disasters.