5 Landslide hazard analysis using the artificial neural network
Before running the artificial neural network program, the training site should be selected. So, the landslide-prone (occurrence) area and the landslide-not-prone area were selected as training sites. Cells from each of the two classes were randomly selected as training cells, with 327 cells denoting areas where landslide did not occur or occurred. First, areas where the landslide did not occur were classified as “areas not prone to landslide” and areas where landslide was known to exist were
assigned to an “areas prone to landslide” training set.
The back-propagation algorithm was then applied to calculate the weights between the input layer and the hidden layer, and between the hidden layer and the output layer, by modifying the number of hidden node and the learning rate. Three-layered feed-forward network was implemented using the MATLAB software package. Here, “feed-forward” denotes that the interconnections between the layers propagate forward to the next layer. The number of hidden layers and the number of nodes in a hidden layer required for a particular classification problem are not easy to deduce. In this study, a 9
× 19 × 2 structure was selected for the network, with input data normalized in the range of 0.1–0.9. The nominal and interval class group data were converted to continuous values ranging between 0.1 and 0.9. Therefore, the continuous values were not ordinal data, but nominal data, and the numbers denote the classification of the input data.
The learning rate was set to 0.01, and the initial weights were randomly selected to values between 0.1 and 0.3. The weights calculated from 10 test cases were compared to determine whether the variation in the final weights was dependent on the selection of the initial weights. The
back-propagation algorithm was used to minimize the error between the predicted output values and the calculated output values. The algorithm propagated the error backwards, and iteratively adjusted the weights. The number of epochs was set to 2000, and the root mean square error (RMSE) value used for the stopping criterion was set to 0.01. Most of the training data sets met the 0.01 RMSE goal. However, if the RMSE value was not achieved, then the maximum number of iterations was terminated at 2000 epochs. When the latter case occurred, the maximum RMSE value was 0.213. The final weights between layers acquired during training of the neural network and the contribution or importance of each of the nine factors used to predict landslide hazard are shown in Table 2.
For easy interpretation, the average values were calculated, and these values were divided by the average of the weights of some factor that had a minimum value. The land use value was the minimum value, 1.00, and the slope value was the maximum value, 3.123. Finally, the weights were applied to the entire study area, and the landslide hazard map was created (Fig. 4). The values were classified by equal areas and grouped into four classes for visual interpretation. The possibility was classified into four classes (highest 10%, second 10%, third 20% and reminding 60%) based on area for visual and easy interpretation. The minimum value is 0.0121 and maximum value is 0.9976. The mean value is 0.3945 and the standard deviation value is 0.3060.