Naoji Koizumi

Changes in groundwater levels or pressures associated with the 2004 earthquake off the west coast of northern Sumatra (M9.0)

Associated with the 2004 earthquake off the west coast of northern Sumatra, changes in groundwater levels or pressures were observed at many observation stations in Japan which are more than 5000 km from the hypocenter. At 38 of the 45 observation stations, there were changes in groundwater levels or pressures. At the 10 observation stations in which the Ishii-type borehole strain instruments were established, changes in crustal strains were also observed. A major part of the changes in crustal strains and groundwater levels or pressures were dynamic oscillations due to a seismic wave. At some stations, coseismic or postseismic rises or drops were also observed. At five stations where both crustal stain and groundwater levels or pressures were observed, postseismic changes in groundwater levels or pressures were consistent with coseismic static steps in crustal strains. At the other five stations, postseismic changes in groundwater levels or pressures did not agree with the coseismic static steps. At two stations of these five stations, it is anticipated that the pore water pressure change in each aquifer locally occurred independently of the change in crustal strain. At another station, postseismic changes in groundwater level possessed the same characteristics as a model removing the temporary deposition. At the last two stations, the causes of the changes are unknown.

Preseismic changes in groundwater level and volumetric strain associated with earthquake swarms off the east coast of the Izu Peninsula, Japan

We have observed the groundwater level at Omuroyama-kita (OMR) well on the Izu Peninsula, Japan since October 1994. During the observation period, there were four earthquake swarms. We detected one or two preseismic groundwater level changes together with preseismic crustal deformation recorded at the other nearby stations. Therefore those preseismic groundwater level changes are considered to be caused by crustal deformation or volumetric strain changes. The crustal deformations are inferred to be caused by the magmatic dike intrusion, which is a possible source of the earthquake swarms off the east coast of the Izu Peninsula.

Earthquake-related Changes in Groundwater Levels at the Dogo Hot Spring, Japan

The Dogo hot spring, situated in Matsuyama City, Ehime Prefecture, Japan, is one of the oldest and most famous hot springs in Japan. The groundwater level or discharge at the spring decreased four times during the past eight or nine Nankai earthquakes. These are large interplate earthquakes that have occurred repeatedly in the western part of the Nankai Trough at intervals of 100–200 years since A.D. 684. To clarify the mechanism of these earthquake-related changes in the water level at the spring, we analyzed groundwater-level data recorded at the spring immediately after the 1946 Nankai earthquake and over the period from 1985 to 2006. We detected the other nine postseismic increases in groundwater level and no decreases, except for a large decrease of 11.4 m related to the 1946 Nankai earthquake. The increases were probably caused by ground-shaking, while the decrease was caused by a change in coseismic volumetric strain. These results lead to the following explanation of the recorded earthquake-related changes in the groundwater level at the Dogo hot spring. Both coseismic changes in volumetric strain and ground-shaking can lead to postseismic changes in groundwater pressure. The increase in groundwater pressure arising from ground-shaking is generally greater than the change in pressure associated with changes in coseismic volumetric strain; however, at the time of the Nankai earthquakes, the spring experiences a large increase in coseismic volumetric strain, leading to a considerably larger decrease in the groundwater level than the increase associated with ground-shaking. Therefore, the groundwater level at the Dogo hot spring usually increases at times of relatively large earthquakes, although the groundwater level or discharge decreases in the case of the Nankai earthquakes.

Water injection experiments and discharge changes at the Nojima Fault in Awaji Island, Japan

Three boreholes (500, 800, and 1800 m) were drilled near the Nojima Fault, which is one of the Hyogoken-Nanbu Earthquake faults. Water injection experiments were conducted from the 1800-m borehole over three periods (1997/2/9–11, 2/12–13, 3/16–25). Groundwater discharge increased at the 800-m borehole during the water injections. The discharge started to increase at between one-half to one full day after the beginning of the injections and decreased substantially just after the stop of the injections. One day after the injections were stopped, their effects on the discharge roughly disappeared. It appeared that these changes in discharge were due to changes in pore water pressure changes induced by the water injections. Therefore, by means of a simple numerical simulation, we analyzed the correlation between the pore water pressure changes and the water injections. In the case of a permeable stratum with a hydraulic diffusivity of 1.0–2.0 km²/day, which is equivalent to 0.08–0.47 darcy on the assumption that the porosity of the stratum is 0.01∼0.1, the calculated pore water pressure changes were highly similar to the observed discharge changes.

Frequency dependence of the groundwater discharge at an artesian well as recognized from tidal fluctuation records

Groundwater discharge, temperature, and electrical conductivity of an artesian well were measured at Nishiawakura, Okayama Prefecture, southwest, Japan. The discharge ranged from 2×10−4 to 5×10−4 m3/s and showed tidal fluctuations of about 5 ×10−5 m3/s induced by strain. Although the groundwater temperature reflected the same tidal fluctuations as the discharge, the electrical conductivity did not manifest tidal fluctuations. The discharge slightly leads the theoretical tidal strain, and its response to the strain of the diurnal constituents is different from that of the semidiurnal constituents. The discharge fluctuations induced by atmospheric loading also have some frequency dependence. These can be explained by the vertical or horizontal groundwater flow(leakage) from the aquifer which has been used in the explanation of the well water level fluctuations induced by tidal strain and atmospheric loading.

Changes in groundwater level associated with the 2003 Tokachi-oki earthquake

Groundwater level and flow rate at 44 wells are continuously observed by the Geological Survey of Japan and the Shizuoka and Gifu Prefectural Governments for monitoring seismic and volcanic activities. The 2003 Tokachi-oki earthquake (M8.0) occurred off the south coast of Hokkaido Island, Japan on September 26, 2003. The epicentral distance to the nearest observation well is about 250 km and that to the farthest is about 1200 km. At the 22 wells, we detected changes in groundwater level or flow rate in relation to the earthquake. Most of the changes are coseismic step-like changes and/or short-period oscillations. In the nearest two observation wells, long-period oscillations with the periods of 39 and 53 minutes were also observed for several days after the earthquake, which is likely due to tsunami. In comparison between distributions of changes in groundwater level and theoretical coseismic strain by the fault model, it is clear that step-like increases were found in the contraction area of the coseismic strain. The relationship between amounts of the observed step-like groundwater-level changes and theoretical ones, calculated by the fault model using strain sensitivities of groundwater level indicates that the groundwater levels in the several wells responded to the coseismic strain.

Groundwater changes associated with the 2004 Niigata-Chuetsu and 2007 Chuetsu-oki earthquakes

The Geological Survey of Japan, AIST, has been monitoring groundwater in and around the Kinki and Tokai districts for earthquake prediction research. The Niigata Prefectural Office has also been observing groundwater for monitoring land subsidence in Niigata Prefecture. The 2004 Niigata-Chuetsu (MJMA 6.8) and 2007 Chuetsuoki (MJMA 6.8) earthquakes occurred in Niigata Prefecture, Japan, on October 23, 2004 and July 16, 2007, respectively. The two earthquakes have a similar magnitude, epicenter, and mechanism. At many of the observation wells, we detected changes in groundwater level or pressure related to the two earthquakes, but no clear precursory changes. At all of our observation wells in Niigata Prefecture, trend changes were observed after coseismic step-like changes for both of the earthquakes. At some of the stations in and around the Kinki and Tokai districts, coseismic trend changes and/or step-like changes were observed. The pattern of the changes were almost similar for the two earthquakes. Those changes were considered to be caused not by the static crustal deformation but by the ground shaking.

A study on the mechanism of coseismic groundwater changes: Interpretation by a groundwater model composed of multiple aquifers with different strain responses

In order to research the coseismic and postseismic groundwater changes for a groundwater system consisting of multiple aquifers we studied the mechanism of coseismic groundwater changes at Yudani hot spring, which is an artesian well supplied by at least three aquifers, in Tottori Prefecture, Japan. We observed the changes in discharge and water temperature at this site in response to earthquakes, tides, and atmospheric pressure changes. For the groundwater system of Yudani hot spring a realistic model composed of three aquifers having different frequency dependencies of groundwater responses to crustal strain and atmospheric pressure was adopted. The tidal and barometric responses of the discharges from the three aquifers, which were cyclic responses over a limited frequency band, were calculated, and the physical parameters of the three aquifers were estimated by fitting a theoretical model to the observed responses. Then, using the strain response over the whole frequency determined by the estimated physical parameters, the theoretical responses of the discharges from the three aquifers to step-like volumetric strain changes were calculated. As a result, the three-aquifer model with different frequency dependencies quantitatively explained the observed groundwater changes at Yudani hot spring due to earthquakes. In addition, the values of the coseismic volumetric strain steps were estimated based on groundwater changes at Yudani hot spring. These values did not agree with the theoretical coseismic volumetric strain changes at Yudani hot spring due to the earthquake fault slips, which were calculated under the assumption that the Earths crust is a homogeneous elastic half-space medium. It is possible that a local strain release, such as aseismic slip at a fault near Yudani hot spring, might have caused the large strain changes estimated from the coseismic groundwater changes.

Effects of seismic ground motion and geological setting on the coseismic groundwater level changes caused by the 1999 Chi-Chi earthquake, Taiwan

The groundwater level changes induced by the 1999 Chi-Chi earthquake were well recorded at the monitoring wells in and around the Choshui River alluvial fan, Taiwan, which is adjacent to the focal region. We analyzed the coseismic groundwater level changes related to the geological setting and seismic ground motion. In a typical fan area, the groundwater levels coseismically rose and those amplitudes increased as the ground acceleration and hydraulic conductivity became larger. In the slope area near the earthquake fault, the groundwater levels coseismically dropped and those amplitudes increased as the ground acceleration became larger. The liquefaction and permeability enhancement, whose degrees depend on the geological setting and seismic ground motion, might explain the characteristics of the coseismic groundwater level changes in the Choshui River alluvial fan.

Recent hydrological and geochemical research for earthquake prediction in Japan

Hydrological and geochemical studies for earthquake prediction in Japan during the last two decades are reviewed. Following the 1995 Hyogo-ken Nanbu (Kobe) earthquake, the central approach to research on earthquake prediction was modified. Instead of precursory detection, emphasis was placed on understanding the entire earthquake cycle. Moreover, the prediction program for the anticipated Tokai earthquake was revised in 2003 to include the detection of preslip-related precursors. These changes included the promotion of the following hydrological and geochemical studies for earthquake prediction: (1) development and/or application of statistical methods to extract small fluctuations from hydrological/geochemical data, (2) evaluation of the detectability of preslip-related anomalies in terms of groundwater levels in wells in the Tokai region, and (3) establishment of a new groundwater and borehole strain observation network for Nankai and Tonankai earthquake prediction research. The following basic geochemical studies were carried out: (1) development of a new monitoring system using a quadrupole mass spectrometer, (2) experimental studies on hydrogen generation by the grinding of rock and crystal powders, (3) comprehensive monitoring of groundwater gas and precise crustal deformation, and (4) mantle-derivative helium observation to compare with seismic velocity structures and the distribution of non-volcanic tremors. Moreover, hydrological and geochemical investigations related to the evolution of fault zones were introduced within the framework of fault zone drilling projects.