Sensitivity analysis of factors controlling earth fissures due to excessive groundwater pumping
Y. Li, P. Teatini, C. Zoccarato
Dept. of Civil, Environmental and Architectural Engineering,
University of Padova, Padova, Italy
N. Friedman, A. Benczur
Informatics Laboratory, Institute for Computer Science and Control (SZTAKI), Budapest, Hungary
S. Ye
School of Earth Sciences and Engineering, Nanjing University, Nanjing, China
L. Zhu
College of Resource Environment and Tourism, Capital Normal University, Beijing, China
Aseisimic earth fissures are complex consequences of groundwater withdrawal and natural hydrogeologic conditions. This
paper aims to improve the understanding of the mechanism of earth fissuring and investigate the relative importance of
various factors to fissure activity, including bedrock geometry, piezometric depletion, compressibility and thickness of the
exploited aquifer. For these purposes, a test case characterized by an impermeable and incompressible rock ridge in a
subsiding basin is developed, where stress/displacement analyses and fissure state are predicted using an interface-finite
element model. Three different methods for global sensitivity analysis are used to quantify the extent of the fissure opening
to the aforementioned factors. The conventional sampling based Sobol' sensitivity analysis is compared to two surrogate
based methods, the general polynomial chaos expansion based Sobol' analysis and a feature importance evaluation of a
gradient boosting decision tree model. Numerical results indicate that earth fissure is forming in response to tensile stress
accumulation above the ridge associated to pore-pressure depletion, inducing the fissure opening at land surface with
further downward propagation. Sensitivity analysis highlights that the geometry of bedrock ridge is the most influential
feature. Specifically, the fissure grows more when the ridge is steeper and closer to the land surface. Pore pressure depletion
is a secondary feature and required to reach a certain threshold to activate the fissure. As for this specific application, the
gradient boosting tree is the most suitable method for its better performance in capturing fissure characteristics.
