Loading...
Geospatial and logistic regression assessment of factors contributing to backwards erosion piping
Semmens, Stephen
Semmens, Stephen
Citations
Altmetric:
Advisor
Editor
Date
Date Issued
2020
Date Submitted
Collections
Research Projects
Organizational Units
Journal Issue
Embargo Expires
Abstract
Backwards erosion piping (BEP) is a form of internal erosion which frequently undermines dams, levees, and other water control structures by removing material from the foundation. The hazard manifests within specific ground conditions common in many alluvial and deltaic environments. Identifying and predicting BEP is difficult because the process is spatially rare and occurs beneath the ground surface. This dissertation presents a body of work contributing an improvement to the forecasting of BEP along levee structures. Three studies are presented assessing BEP occurrence across the Lower Mississippi Valley of the United States and the IJssel and Waal rivers of the Netherlands. Two logistic regression analyses and a stepped empirical assessment are performed. The purpose of these studies is twofold. First, assess a series of factors suspected of or previously identified as predictors of BEP. Second, generate improved BEP prediction models. Several results of note may be drawn from this work.1. Land cover and certain geologic deposits are likely less important predictive parameters than previously believed.2. Critical gradient, exit gradient, and blanket thickness exhibit consistent correlation and predictive significance for BEP.3. Adverse orientation of local geologic deposits and factor of safety show promise as predictive factors but require additional investigation for effective application.4.For the first time, blanket thickness is shown to influence the likelihood of BEP independently of factor of safety. This indicates loading conditions alone may not be able to account for BEP occurrence.5.An empirically derived Naive Bayes model is presented. The model is designed to generate probabilities of BEP prediction based on blanket thickness and factor of safety.6.A logit model is presented for widespread BEP forecasting. The model outperforms previous attempts while using only three regionally-independent factors: critical gradient, exit gradient, and thickness sum of clay within the blanket.Together, these insights may be applied to improve BEP hazard assessment along levee structures.
Associated Publications
Rights
Copyright of the original work is retained by the author.