Permanent gully caused by high-intensity agricultural activities and water erosion have severely degraded croplands in the Northeast China, a global critical black soil regions. The precise differential characteristics of the primary factors contributing to the formation of permanent gully across diverse regions and at varying scales remain unclear. It is necessary to select representative regions and scale effect models to reveal dominant factors and their influencing scales. This study selected certain permanent gully influencing factors and built three models (CF, ME, and CNN) to identify dominant controlling factors of permanent gully and assess susceptibility. ME and CNN models revealed drainage density and distance to rivers as primary drivers, distinguishing this region from others. These factors represent the degree of hydrological connectivity and energy accumulation, which trigger the occurrence and development of permanent gully when reaching a threshold value. CF model highlighted areas within specific distance to rivers and drainage density threshold are more susceptible to gully erosion. Evaluation using ROC curves showed the CNN model outperformed ME model (AUC = 0.83) and CF model (AUC = 0.77), attributed to its ability to integrate multi-scale environmental factors. The gully erosion susceptibility mapping also revealed high-risk areas in the watershed, which were mostly found in the drainage density of 0.42-0.51 km/km2 and distance of 2244 m to rivers. These research results assist in preventing permanent gully to protect land resources health and underscore the importance of prioritizing high-risk areas for conservation efforts, such as implementing vegetation buffers to mitigate the development of permanent gully.