EARTHQUAKES AND HEAT FLOW

Heat flow is an important physical parameter for interpreting tectonic processes such as the frictional resistance to the relative motion between the lithospheric plates. However, large earthquakes can release a great amount of groundwater that may affect heat flow; thus it is important to understand how earthquake-released groundwater may affect heat flow in order to effectively use the latter to interpret plate-boundary processes. In this study we use temperature data from deep monitory wells in Taiwan before and after the 1999 M_w7.5 Chi-Chi earthquake to show that large earthquakes may significantly affect heat flow, not only shortly after the earthquakes but also throughout most of the inter-seismic period.

Following the Chi-Chi earthquake, most deep monitory wells in the alluvial fan beyond the thrust front showed an increase in groundwater temperature and geothermal gradient. On the other hand, most wells in the elevated terrains near the thrust front showed a decrease in groundwater temperature and geothermal gradient. Interpretation of these changes requires an understanding of the mechanism by which earthquake releases groundwater because the mechanism is closely related to the flow paths of the released groundwater and thus its effect on heat flow. Several hypotheses were proposed, including expulsion of deep crustal fluids due to the static elastic strain, consolidation or liquefaction of near-surface deposits, and changes in near-surface permeability, but none satisfied all the existing evidences. Wang et al. (2004) showed that large earthquakes may change the state of anisotropy of permeability of the hydraulic system near the epicenter of an earthquake and that this may explain all the relevant observations.

Taiwan is a mountain belt formed by the collision between the Philippine Sea plate and the Eurasian plate. The fold-and-thrust foothills are underlain by alternating sandstone and shale formations, with an effective layer-parallel permeability several orders of magnitude greater than that normal to the layers. Following the Chi-Chi earthquake, numerous subvertical tensile cracks appeared in the foothills, which breached the impervious beds in the sedimentary sequence and greatly enhanced the vertical permeability of the hydrological system, allowing downward draining of groundwater. Well data show that after the Chi-Chi earthquake the vertical hydraulic diffusivity increased to at least 3 m²/s at depths exceeding 250 m. Many wells in the foothills above the thrust fault dropped in water level, and a tunnel beneath the foothills experienced sudden downpour right after the earthquake. The predicted time-dependent change in the earthquake-induced streamflow based on this model also agrees closely with observation. Finally, most monitory wells on the alluvial fan beyond the thrust front showed a rise in the groundwater level after the earthquake, indicating influxes of groundwater. Lowering of the groundwater level in mountainous terrain and increased streamflow were also reported after the 1989 Loma Prieta earthquake in California and the 1995 Kobe earthquake in Japan near the respective epicenters; thus the model of earthquake-enhanced vertical permeability may apply to different plate boundaries.

Observation shows that the post-seismic vertical permeability returns to the pre-seismic value within about 2 years. Thus the timescale for the post-seismic recovery of permeability is much shorter than the recurrence interval of large earthquakes. Between large earthquakes, the mountain slopes will again be recharged with groundwater, only to be released during the next large earthquake. The repeated downward draining of groundwater from the mountains during large earthquakes is effective in flushing the geothermal heat in the foothills and discharging it to the nearby lowlands – explaining the temperature data from deep wells in Taiwan after the Chi-Chi earthquake.

The post-seismic heat flow may slowly recover after the earthquake. The time required for this recovery depends on the thermal conductivity of rocks and the depth where a significant portion of the geothermal heat is flushed away. Using the well data in Taiwan for the vertical permeability and the depth of groundwater penetration after the Chi-Chi earthquake, we show that a significant portion of the geothermal heat in the Taiwan foothills may have been flushed down to a minimum depth of 100 m. Given a reasonable range of thermal diffusivity from 10^-6 to 10^-7 m²/s for rocks and sediments, the characteristic time for the heat flow to recover to the pre-seismic level is from 300 to 3000 yr. Recalling that this characteristic time is comparable to, or longer than, the recurrent interval of large earthquakes, we conclude that large earthquakes in Taiwan may significantly alter the regional heat flow, not only shortly following the earthquakes but also throughout much of the inter-seismic interval. The latter conclusion is consistent with the existing heat flow data in Taiwan.

In other parts of the world, where large earthquakes are frequent, similar earthquake-induced changes of heat flow may occur. Thus it is important to account for the earthquake effect before using heat-flow data to interpret tectonic processes.