Monitoring long-term gully erosion and topographic thresholds in the marginal zone of the Chinese Loess Plateau

•A multi-scale gully erosion monitoring method was developed.•Long-term gully morphology changes from 1959 to 2018 were monitored.•Impact of land use and topographic factors on gully-head retreat was evaluated.•Topographic thresholds of gully initiation were determined. Gully erosion has been widely regarded as an extreme process of land degradation around the world. Gully development rate and topographic thresholds obtained by measuring the morphological parameters of multiple gullies and their influencing factors are important indicators used to characterize gully activity and potential locations of gully initiation. However, existing gully measurement and observation methods are time-consuming and labor-intensive, and usually only a single or several gullies can be measured at a time. The objectives of this study are to develop a multi-scale gully erosion monitoring method by combining unmanned aerial vehicle (UAV) technology with historical aerial photogrammetry and a field survey, achieve the measurement of gully morphological parameters (length, width, depth, area, width-depth ratio, cross-sectional area and volume) and environmental factors (land use, slope gradient and drainage basin area) for numerous gullies in larger regions, and finally, solve the issue of the accurate calculation of regional topographic thresholds and gully development rates. We applied a multi-scale gully erosion monitoring method in the marginal zone of the Chinese Loess Plateau to investigate the gully changes from 1959 to 2018 and further identified long-term gully development patterns and their main control factors. We identified 46 permanent gullies via a field survey. Long-term developments in 30 permanent gullies were obtained using historical aerial photographs and UAV images. The topographic threshold relationship between the upslope drainage basin area (A) and local slope gradient (S) was determined using the remaining 16 incipient gullies derived from UAV images. The results indicated that the gully dimensions experienced significant expansion; the linear gully-head retreat was considerably stronger than the gully widening and deepening. The contribution of the eroded volume from the gully-head retreat to the total eroded volume was 6.1 times that of the gully sidewall retreat, indicating that gully-head retreat was the primary mechanism for gully development during the study period. The area of the gully-head drainage basin played a more significant role than the