Land subsidence caused by seasonal groundwater level fluctuations in Kawajima (Japan) and one-dimensional numerical modeling with an evolutionary algorithm

Abstract

The causes of land subsidence in Kawajima, Japan, have been investigated through data compilation and numerical modeling. Land subsidence has progressed despite a gradual increase in the hydraulic head in the long term. Taking into account the temporal changes and depth distribution of groundwater abstractions, the contraction of formations, and the complexity of the hydrogeological structures, it is proposed that agricultural groundwater use is one of the main triggers for land subsidence. A one-dimensional numerical simulator for coupled groundwater flow and soil deformation was developed with an evolutionary algorithm for model calibration. The calculated spatiotemporal changes in the past-maximum effective stress showed that plastic consolidation in the clayey layers progressed part by part every summer season resulting in long-term and gradual land subsidence under the same range of groundwater level fluctuations. The results also showed that the plastic deformation occurred in both the Holocene and Pleistocene sediments in the drought years, leading to significant subsidence. The model’s predictive performance showed good potential except for a structural prediction error after the Tohoku Earthquake in 2011. The scenario analysis indicated that management of the groundwater level in summer is one of the effective countermeasures in suppressing land subsidence caused by seasonal groundwater level fluctuations. These methodologies and findings can be used for groundwater management in similar cases around the world. Additional investigation is necessary on the influence of large earthquakes in deformation conditions in order to further improve the developed model.