Earthquake-induced landslide prediction using back-propagation type artificial neural network: case study in northern Iran

Landslides can cause extensive damage, particularly those triggered by earthquakes. The current study used back propagation of an artificial neural network (ANN) to conduct risk studies on landslides in the area affected by the Manjil-Rudbar earthquake in Iran in 1990 (M = 7.7). Newmark displacement analysis was used to develop an earthquake-induced landslide hazard map for the blocks representing Chahar-Mahal and Chalkasar near the earthquake epicenter, an area of 309 km(2). The input data included soil cohesion, soil friction angle, unit weight of soil, unit weight of water, distance from hypocenter, slope, and earthquake magnitude as effective parameters for landslide occurrence. The hazard map was compared with an inventory map and other research findings. The results indicated that the landslides predicted by ANN covered 50% of the inventory map of the study area (2088 of 4097 slide cells). The results of the current study suggest that the ANN method is relatively efficient for accurate prediction of landslides.

Automated summary

This study used an artificial neural network to conduct risk studies on landslides in the area affected by the Manjil-Rudbar earthquake in Iran in 1990. The results showed that the ANN method is relatively efficient for accurate prediction of landslides, covering 50% of the inventory map of the study area. The hazard map developed through Newmark displacement analysis was compared with other research findings, highlighting the effectiveness of the ANN approach.