Analyzing Building Damages of a Major Earthquake Using GRASS and R
Presentation | Presented
- Wen-Chieh (Jeffrey) Wang, Chaoyang University of Technology
R is a free software environment for statistical computing and graphics. Due to its well-developed capabilities of handling the common autocorrelation issue in spatial data, R has become a popular companion to GRASS in conducting spatial data analysis. This presentation demonstrates a GIS application that uses GRASS with R as spatial data analysis tools to examine the spatial aspect of building damages in a major earthquake.
On September 21, 1999 an earthquake measured 7.3 on the Richter scale occurred in central Taiwan. It damaged more than seven thousand buildings. A collaborative effort among government agencies and universities at that time had collected valuable dataset regarding those damaged buildings in time. This study first uses GRASS to visually explore the spatial distribution pattern of the damaged buildings in the dataset. It then divides the study region into smaller spatial units and calculates the number of damaged buildings per unit area that will serve as the dependent variable of the spatial regression in the later analysis process. One set of the independent variables this study uses for regression is the peak ground acceleration (PGA) data originally provided by the Central Weather Bureau in Taiwan. However, the study has to apply some GRASS interpolation functions to the actual measured values at the 384 earthquake monitoring stations around the island beforehand in order to obtain the PGA value at the centroid of each spatial unit for regression. Finally all prepared variables are brought into R for further statistical data visualization, exploration, and modeling.