Avalanche occurrence represents one of the most significant natural hazards in mountainous regions, capable of inflicting substantial damage on transportation infrastructure, tourism activities, and human life. Consequently, the identification of susceptible zones and the accurate prediction of this phenomenon are of paramount importance for risk management and safety planning. The aim of this study is to conduct hazard zonation for avalanche occurrence on the Sarein to Alvares Ski Resort road using a deep learning approach based on a Deep Convolutional Neural Network (DCNN). To this end, ten environmental factors including elevation, slope, aspect, Topographic Position Index (TPI), Terrain Ruggedness Index (TRI), snow cover, wind speed, vegetation cover, land use/land cover (LULC), and distance from the road were selected as input variables. Following preprocessing, these factors were incorporated into the model as informational layers. Avalanche occurrence data were collected through field studies and GPS surveying, and the model was trained in the Python programming environment. The results indicated that the model achieved stable convergence in fewer than eight epochs, with a minimum validation loss recorded as 0.0191. The model accuracy increased from approximately 80% at the beginning of training to 98.03% during the validation phase. Furthermore, the precision values ranged from 0.95 to 0.99, and recall values ranged from 0.98 to 1.00, indicating the model's high capability in identifying high-risk zones. An AUC value of 0.985 further demonstrated the model's very high discriminatory power among hazard classes. The final results indicate that the DCNN model is an efficient tool for avalanche hazard analysis and prediction and can be used in crisis management and safety planning in mountainous regions.