Permanent water level drop associated with the Spitak Earthquake: observations at Lisi Borehole (Republic of Georgia) and modelling

We present the results of a series of measurements that were made between 1988 and 1992 at Lisi (Georgia). Water level variations in the Lisi well, barometric pressure, precipitation (including rain and snow) and temperature measurements were made during this period. A hydraulic ‘slug test’ has been performed more recently in the well. Two major seismic events occurred during the observation period in the Caucasus area. The Spitak seismic event of 1988 December 9, 110 km from the Lisi borehole, left a clear post‐seismic hydraulic signature, whereas the second event, that of 1991 April 29, 125 km from the borehole, did not seem to induce any detectable anomaly. First, we analyse the tidal and barometric responses of the water level in order to calibrate the borehole and to determine the hydraulic parameters of the aquifer. Then we develop a model for aquifer recharge by meteoric precipitation. Finally, we compare our model with the observed water level variations in the well. We highlight anomalous behaviour that correlates with the earthquake, with the following characteristics: the water level drops about 75 cm with a time constant of 6.6 days. The initial water level is never recovered and the change appears permanent on the scale of the period of observation. Since it is delayed in time, the anomalous water level cannot be attributed to coseismic deformation. Following the suggestion of some authors that the observed behaviour could be related to damage of the aquifer due to the passage of seismic waves, we attempted to take this process into account and to model the resulting water level variations in the aquifer. A double porosity model (including fractures and a porous medium) has been used to describe the modifications undergone by the medium. The medium is discretized at two different scales—(1) at the scale of a porous block and (2) at the scale of the fractured system (which may include a finite number of porous blocks). Using this basic model we have tested three interpretative models: (1) a variable‐permeability model; (2) a locally varying porosity model; (3) a model combining (1) and (2). We show that an increase in permeability by a factor of 2 is needed in order to describe the permanent water level drop, whilst the locally varying porosity zones account for the observed time constants of the anomaly. Using this model, simulations show that we are able to describe water level variation associated with precipitation for the whole period of observation.