Long-Term Creep Rates on the Hayward Fault: Evidence for Controls on the Size and Frequency of Large Earthquakes

earthquakes with short recurrence times (161 65 yr), probably kept short by a 26%–78 % aseismic release rate (including postseismic). Its interseismic release rate varies locally over time, as we infer from many decades of surface creep data. Earliest estimates of creep rate, primarily from infrequent surveys of offset cultural features, revealed distinct spatial variation in rates along the fault, but no detectable temporal variation. Since the 1989 Mw 6.9 Loma Prieta earthquake (LPE), monitoring on 32 alinement arrays and 5 creepmeters has greatly improved the spatial and temporal resolution of creep rate. We now identify significant temporal variations, mostly associated with local and regional earthquakes. The largest rate change was a 6-yr ces-sation of creep along a 5-km length near the south end of the HF, attributed to a regional stress drop from the LPE, ending in 1996 with a 2-cm creep event. North of there near Union City starting in 1991, rates apparently increased by 25 % above pre-LPE levels on a 16-km-long reach of the fault. Near Oakland in 2007 anMw 4.2 earthquake initiated a 1–2 cm creep event extending 10–15 km along the fault. Using new better-constrained long-term creep rates, we updated earlier estimates of depth to locking along theHF. The locking depths outline a single,∼50-km-long locked or retarded patchwith the potential for anMw ∼ 6:8 event equaling the 1868 HF earthquake. We propose that this inferred patch regulates the size and frequency of large earthquakes on HF. Online Material: 2007 event creep models, plots of creepmeter data, maps and cross-sections of relocated microearthquakes and active fault traces, and iterative solu-tions for depth of creep.