Potential Evaluation of Forest Road Trench Failure in a Mountainous Forest, Northern Iran

After road construction in steep and mountainous areas, there is always a risk for trench failure. Estimation of this probability before forest design urgent. Besides, to decrease failures costs risks, it necessary classify their occurrence probabilities identify the factors affecting them. The present study compares three statistical models logistic regression, frequency ratio, maximum entropy. robust one was applied generate susceptibility map roads two watersheds Northern Iran. Also, all repairing were estimated, subsequently, existing surveyed area, detecting 844 failures. Among recorded failures, 591 random cases (70%) used modeling, others (30%) as validation data. digital layers, including failure locations, prepared. Three maps simulated using outputs mentioned methods GIS environment. resulted combined with repair cost prices analyzed statistically evaluate unit per meter square results showed that regression model had an Area Under Curve (AUC) 74.6% identifying failure-sensitive areas. probabilistic ratio Entropy 68.2 65.5% accuracy, respectively. Based on model, distance faults terrain slope highest effects trenches According result, about 43.25% network located »high« »very high« risky estimated regulating profiling ditches along approximately 108,772 $/km.