(FIG-15)
So the conclusion of the first part is this： This is the fault zone here, and sometimes there may be fluid and sometimes not. If there is no fluid in the fault zone and if the magnitude gets up to 6 or so, then temperature will go up to 1000 degrees and melting happens. Once melting happens, friction drops, and so in a way the whole thing can operate at relatively low stress, because melting reduces friction, and also if the friction is reduced, the whole thing can slip very easily. The melting would promote very rapid abrupt slippage. So this means in a way you get into a runaway situation. Before melting it wants to slip but there is resistance but, as soon as melting starts, you get into a runaway situation - the whole thing will go, and this is actually the explanation for that gap in the magnitude-frequency relationship - I'll get back to that.
However, if there is fluid, the situation can be different. Before the temperature goes up to 1000 degrees, the fluid is heated, and even at high pressures and temperatures, if fluid is heated it expands, and if it expands, the pore pressure increases, thereby lubricating the surface. So again for large events, if the fluid is present, the friction will drop, and the same thing will happen. What happens in the real world can be extremely complicated because there will be variations in compressibility, permeability, porosity, amount of fluids, strength, stress, heterogeneity and everything.
One important parameter here is permeability. Just to the first order, if the permeability is less than 10-18 m2, basically water can't escape - it is not permeable enough. So water will stay there and lubrication happens. However, if permeability is very large, the water will escape from the fault zone so this mechanism is not going to work. So the permeability measurement is very important, and again down-hole measurements will be important.
So to conclude for the first part, basically what I'm going to say is： the behaviour of seismic sequence around subduction zones - aseismic or seismic - can be really controlled by the slip itself. If the magnitude gets bigger, the thermal process takes over, so the slip behaviour can change completely. It's a non-linear process, so even a small perturbation in the initial condition can produce very different results.
This kind of gap in the magnitude-frequency relationship is consistent with this kind of mechanism. Up to magnitude 6 or something, fault wants to slip, and it can slip, but there is a fair amount of resistance because of friction. But once magnitude gets up to 6 to 7, because of heating in the fault zone, the friction drops which promotes very rapid sliding - brittle sliding, and it can't stop. Once it exceeds the threshold the slip can't stop, and it can go all the way as far as it can go. In the case of the Nankai Trough it's about 8.2 to 8.4; in case of the San Andreas it's magnitude 8. It's a very simple mechanism. OK, this is the conclusion.
For the remaining part of my talk I want to elaborate on some detail why this is consistent with whatever data we now have in seismology.
(FIG-16)
Well, the reason why we came up with this kind of idea is this： About a year ago we looked at a deep focus earthquake in Bolivia. This is not a shallow earthquake; this earthquake happened at a depth of about 600 km beneath Bolivia and the magnitude is 8. For this event, quality of seismic data was very high, so we could make very detailed studies. I'm not going to talk about all the details. Bottom line conclusion is this： In terms of energy budget, we could determine the minimum potential energy; that is about 1.4 x 1018 joule. Also you can measure radiated energy by using seismology, and that is approximately 5 x 1016 joule which is almost nothing.
