Publication Date : 2016-03-13
Author : dela Cerna, M. A.Maravillas, E. A.
Disaster Management Theme :
Disaster Type : Landslide
Document Type : Research Paper
Languange : en
Link : http://www.iaeng.org/publication/IMECS2016/IMECS2016_pp25-30.pdf
This paper presents how spatial data mining is achieved using spatial clustering. The study utilized partitive clustering technique such as K-Means and Self-Organizing Map Algorithms in Artificial Neural Network to map landslide hazard areas using the datasets of eight causative landslide indicators such as slope gradient, vertical displacement, drainage density, the rate of weathering, lithology, ground stability, soil type, and vegetation. Each technique has limitations and concentrates only on particular kind of computation. Based on the result, vertical displacement and slope gradient are the two major aspects that trigger the landslide and dominate in all factors. With these techniques, landslide hazard zone are identified. Hence, the model can map the areas under study and can be a tool to visualize the identified causative factors of landslides.