Statistical modeling of water pipeline damage in earthquakes

Date
2017
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
A large dataset of water pipeline damage from the February and June 2011 earthquakes in Christchurch, New Zealand is used to fit four mathematical model types—logit, boosted regression trees (BRT), and random forest (RF), and the repair rate (RR) method common in the literature. Cross validation and holdout validation are used with multiple metrics to fully evaluate the models’ ability to accurately predict the total number and approximate spatial distribution of damaged pipes; to correctly classify each individual pipe as damaged or not, and to describe the relative importance of pipe and earthquake attributes in predicting damage. Results suggest that while BRT offers the best overall performance, logit offers the advantages of a closed form solution and an ability to compare pipe materials explicitly, and the far simpler RR method is very good at predicting the total number of damaged pipes, though less capable of prediction at the individual pipe or suburb level. The analysis provides evidence that “modified” PVC (MPVC), UPVC, Polyethlyne 80B (PE80B), High Density Polyethlyne (HDPE), and Cast Iron (CI) were associated with the least damage, and Galvanized Iron (GI) with the most; and that the more recent the type of trench it is in, the less likely a pipe is to be damaged, even when controlling for the pipe age. The analysis highlights the need to compute and report the predictive errors of different types and acknowledge them in using the models for subsequent analysis.
Description
Keywords
Applied sciences, Earth sciences, Boosted regression tree, Earthquake, Logit, Random forest, Statistical model, Water pipeline
Citation