Extraction of hydrological anomalies related to earthquakes

Abstract

Time series analysis using state space modeling is applied to extract hydrological anomalies related to earthquakes. This method can break down the observed ground-water level and/or groundwater discharge into four components: (1) barometric response, (2) response to earth tide, (3) observation noise component and (4) ‘residual water level’, which means gradually changing trend of the observed data. Groundwater level observed at Haibara observation well, Shizuoka Prefecture, central Japan is analyzed by this method. In the groundwater level at the Haibara well, 28 coseismic changes are detected during the period from April 1981 to December 1997. There is threshold in the relationship between magnitude and hypocentral distance of the earthquakes, which affect the coseismic changes in the residual water level. All of the coseismic changes of the water level at Haibara are decreasing, however, 33% of estimated coseismic step of volumetric strain is contraction. This result shows the whole amount of the residual water-level change at the Haibara well is not explained by the hypothesis that the water level of confined wells changes in response to volumetric strain. Four possible pre-earthquake changes are detected in the residual water level at Haibara well. Finally, we show the relationship between this method and the TANK model, one of the conceptual runoff models, in the estimation of the response to precipitation.