What drives gullies in Spain’s olive landscapes? A regional analysis of gully activity

• 475 gully heads identified and classified in olive groves of Guadalquivir basin. • Gully activity classified into 3 types: stable, newly formed, and active. • GHI model predicts gully initiation well; AUC reaches 0.93. • GHI outperforms topographic threshold model for all gully types. Gully erosion poses a significant threat to Mediterranean basins. However, precise spatial data on gullies location at regional scale is scarce. Current models often use the topographic threshold index, TT, which works locally and predicts gully location based on slope and drainage area but does not distinguish between actively retreating and stabilized gullies. We aim to close this gap by (i) constructing a dataset of gully head location and activity in olive groves in southern Spain across different landscape regions and (ii) evaluate a process-oriented model to distinguish gully head presence and activity based on the Gully Head Initiation (GHI) index. This GHI integrates slope, drainage area, precipitation, curve number (CN), soil type, and clay content to predict gully head formation. Photointerpretation of four 25 km 2 study areas between 2008 and 2019, 3 gully head activity classes were identified, with 261 existing active, 76 new active and 138 stable gully heads identified. The GHI clearly differentiates gully head from non-gully head pixels (AUC = 0.93). Furthermore, it distinguished between existing active and stable gully heads, as well as new active and existing active ones. However, there are no significant differences between stable and new active gully heads. Besides, applying the model separately by landscape type is not as effective as applying it across all landscapes. These results highlight the potential applicability of GHI at the regional scale and confirm that the factors considered in the GHI index affect both gully head initiation processes and erosion dynamics.