Transient deformation signals related to hydrology: Examples, mechanisms, and significance

Roeloffs, Evelyn
US Geological Survey, Vancouver, WA, evelynr@usgs.gov

Transient signals related to hydrology have been recorded by creepmeters, borehole strainmeters, and two-color EDM measurements for a decade or more prior to the widespread implementation of high-resolution space geodesy techniques. Understanding these observations is greatly facilitated by comparing the deformation measurements with records of groundwater levels, surface-water discharge, and precipitation when those are available from nearby sites. In many cases, hydrologic phenomena are unrelated to the tectonic or magmatic processes being studied and are important only because they introduce noise into the measurements. For example, creepmeters can record displacements caused by incremental downslope movement of near-surface materials, usually in response to heavy rainfall on soils that are already very wet ? conditions similar to those well-known to cause landslides. Borehole strainmeters respond to a wide variety of hydrologic sources: the weight of precipitation on the ground surface; discharge changes in nearby rivers; groundwater pressure variations caused by pumping; and the natural fluid pressure variations due to seasonal rainfall and evapotranspiration. It should be noted that fault movement and seismic waves change subsurface fluid pressures and groundwater discharge. In particular, seismic waves can initiate persistent water-level changes as far as hundreds of km from the epicenter. Nearby borehole strainmeters also record signals in some of these cases, raising the issue of whether the signal is caused fundamentally by deformation, or by fluid pressure change. Some data show that fluid-pressure changes from distant earthquake occur at depths of several km, and suggest they may be related to remotely triggered seismicity. Groundwater level variations generally complete a cycle with an approximately annual period, but time histories of pressure change are too complicated to approximate with a small number of harmonic components. Exact repeatability from year to year is the exception rather than the rule. The onset of annual recharge is frequently very abrupt, and so are pressure changes caused by different initiation times for irrigation, as spring weather varies from year to year. Regardless of the cause, abrupt changes are broadband signals that are very difficult to filter from deformation data. Although complete removal of hydrologic "noise" from deformation records is challenging, identification of such signals is facilitated by accumulating a long-term record not only of deformation, but also of hydrologic variations near the site.