Secular Variation in the Storage and Dissipation of Elastic Strain Energy Along the Central Altyn Tagh Fault (86-88.5°E), NW China

Abstract:

In elastic rebound theory, hazard increases as interseismic strain rebuilds after rupture. This model is challenged by the temporal variation in the pacing of major earthquakes that is both predicted by mechanical models and suggested by some long paleoseismic records (e.g., 1-3). However, the extent of such behavior remains unclear due to a lack of long (5-25 ky) records of fault slip. Using Monte Carlo analysis of 11 offset landforms, we determined a 16-ky record of fault slip for the active, left-lateral Altyn Tagh fault, which bounds the NW margin of the Tibetan Plateau. This history reveals a pulse of accelerated slip between 6.4 and 6.0 ka, during which the fault slipped 9 ⁺¹⁴/_-2 m at a rate of 23 ⁺³⁵/_-5 mm/y, or ~3x the 16 ky average of 8.1 ^+1.2/_-0.9mm/y.

These two modes of earthquake behavior suggest temporal variation in the rates of stress storage and release. The simplest explanation for the pulse is a cluster of 2-8 M_w > 7.5 earthquakes. Such supercyclicity has been reported for the Sunda (4) and Cascadia (3) megathrusts, but contrasts with steady slip along the strike-slip Alpine fault (5), for example. A second possibility is that the pulse reflects a single, unusually large rupture. However, this Black Swan event is unlikely: empirical scaling relationships require a Mw 8.2 rupture of the entire 1200-km-long ATF to produce 7 m of average slip. Likewise, Coulomb stress change from rupture on the adjacent North Altyn fault is of modest magnitude and overlap with the ATF. Poor temporal correlation between precipitation and the slip pulse argues against climatically modulated changes in surface loading (lakes/ice) or pore-fluid pressure. "Paleoslip” studies such as this sacrifice the single-event resolution of paleoseismology in exchange for long records that quantify both the timing and magnitude of fault slip averaged over multiple ruptures, and are essential for documenting temporal variations in fault slip as we begin to use calibrated physical models of the earthquake cycle to forecast time-dependent earthquake hazard (e.g., 6,7).

1. Weldon et al., 2004 GSA Today 14, 4; 2. Rockwell et al., 2015, PAGEOPH, 172, 1143; 3. Goldfinger et al., 2013, SRL, 84, 24; 4. Sieh et al., 2008, Science, 322, 1674; 5. Berryman et l., 2012, Science, 336, 1690; 6. Barbot et al., 2012, Science, 336, 707; 7. Field, 2015, BSSA, 105, 544.