Predicting the Probability of Landslide using Artificial Neural Network

Landslide is a natural occurrence. But there are also some manmade reason for landslides. As is it is a natural occurrence, man can't prevent this. But the damages can be mitigated and loss of properties lives by an efficient prediction of this. This study tries to predict the occurrence of landslides in Bangladesh and study area was Chittagong City Corporation area and Cox's bazar. There are many reason for landslides. In the case of five parameters are considered for predicting landslides. These are rainfall data, rain data for previous five days, elevation, slope, soil type. Elevation and slope are collected from DEM of Chittagong City Corporation area and Cox's bazar. Rain data from weather forecast. An artificial neural network model is developed. Initially a weight is assumed then adjust the weight using back-propagation algorithm for getting better prediction result. Train the model using train dataset and evaluate the model using test dataset. The accuracy become highest 93.87% using this model.