(Guest post by Yoshiyuki Tanaka)
In the Frontier Letter published on Earth, Planets and Space by Tanaka et al. (2015) titled “An estimate of tidal and non-tidal modulations of plate subduction speed in the transition zone in the Tokai district,” the authors estimated how large the slip velocity in the transition zone is amplified by long-term sea-level changes for the first time. Demonstration has been made for the Tokai area in Japan, where the mixed effects of the Kuroshio Current and the Pacific Decadal Oscillation (PDO) are remarkable (Qiu and Chen 2010). The result may indicate that large spatial-scale climate changes can be one of the factors to modulate a probability of earthquake occurrences.
Geodetic and seismological observations have revealed that slow earthquakes, including non-volcanic tremors and slow slip events, occur on the plate subduction interface over the world (e.g. Schwartz and Rokosky 2007). The source areas of deep slow earthquakes are distributed in depths of approximately 30—40 km in the transition zone, which is located just below the seismogenic zone where great earthquakes occur. In the transition zone, the fault strength is weaken due to the presence of high-pressure fluids, supplied from the subducted slab. As a result, slow earthquakes are more easily triggered by small external stress disturbances, than ordinary earthquakes.
Previous studies analyzing deep tremors in southwest Japan show that the number of tremors which occur per unit time exponentially increase with increasing tidal stresses. This exponential relationship between tremor rate and stress is consistent with a well-known non-linear rock frictional law, which has been confirmed by laboratory experiments, if we consider tremor rate slip velocity. In other words, increase in tremor rate can be interpreted as acceleration of plate subduction velocity in the transition zone. This acceleration will loosen a part of the locked plate interface at the transition zone. Consequently, the slab loads the seismogenic zone more strongly and enhances seismicity. On the basis of this idea, Ide and Tanaka (2014) showed that variations in the tremor rate predicted from tides are well correlated with long-term seismicity in the Nankai Trough.
If tides can affect long-term seismicity in the seismogenic zone, then, can long-term (e.g. decadal time-scale) sea-level changes also affect seismicity? The answer seems no, at a first glance, because decadal changes in the see level is an order of 1 cm, which is 1-2 order(s) of magnitude smaller than sea level changes caused by tides. However, if the above non-linear relationship in the transition zone is considered, a small stress change can modulate the slip velocity when combined with faster stress changes with larger amplitudes.

Fig. 1 shows this idea. In this example, amplitude of the added slow variation in the right panel is only 1/10 of that of the faster variation, and the two blue curves are hard to discern by eyes. However, the output slip velocity is enhanced by 30% when the superimposed longer-term variation is larger (around t=3 and 15), compared with periods when the longer-term variation is smaller (around t=9).

Fig. 2 shows the results for cases excluding and including the non-tidal variation in the ocean bottom pressure (OBP) in the Tokai area. When excluding the non-tidal effect (blue curve), an 18.6-yr variation becomes dominant in the rate of stress change, which is computed with the estimated slip velocity. When including it (red curve), the temporal pattern in the rate of stress change is dominated by the
variation in long-term variations in OBP (phase becomes opposite; lower OBP unclamps the locked transition zone, which increase stress). The difference between the two cases amount to a few kPa/yr and the latter case correlates well with long-term seismicity. This result indicates that long-term variations in OBP should not be neglected even if their amplitudes are much smaller than those of tides. The decadal variation in OBP originates from changes in wind stress over the central North Pacific. The result may indicate that large spatial-scale climate changes can be one of the factors to modulate a probability of earthquake occurrences.
The data seem to be compelling, the explanation physically plausible. It would be good to replicate this study elsewhere.
Who would have thought it?