Quantitative empirical models were proposed to estimate the
extent of stress influence by using groundwater level variations
triggered by earthquakes. The extent of stress influence was
defined as the distance over which an earthquake can induce a step
change of the groundwater level. The proposed classification system,
used to categorize the groundwater level variations triggered
by earthquakes, provides a detailed record of the dynamic characteristics
of the groundwater responses to earthquakes. Only the
groundwater level variations with a step change were analyzed
for determining the extent of stress influence. The relationships
between the maximum extent of stress influence and the earthquake
magnitudes were derived and analyzed. The maximum
extent of stress influence of a step-type anomaly from groundwater
observation is about 250 km and the minimum detectable
earthquake magnitude is 5.0 in the Taiwan area. The relationship
between point forces and earthquake magnitudes was determined
using the analytical solution from poroelastic theory. A
semi-empirical model (Model I) and a fully-empirical model
(Model II) were proposed to estimate the maximum extent of
stress influence from the step change of the groundwater level
and the earthquake magnitude. An earthquake event was used
for model validation and to demonstrate the potential for estimating
the area of anomalous stress. The results show that the cross
areas of these two models, though, do not cover epicenter, the
areas close to it. Model I shows better results than Model II, in
despite of the validated RMSE of epicentral distance is smaller in
Model II than in Model I. Complex geological structures and material
heterogeneity and anisotropy may explain the disagreement
between the models. Nevertheless, the results show the potential
of using groundwater level variations for capturing seismic information.
The proposed model can be used to preliminarily evaluate
the earthquake effect in hydraulic engineering, mining engineering,
and carbon dioxide sequestration. More sensitive wells construction,
more data collection, and further model refinements
can improve the proposed model for estimating the extent of stress
influence