It's a stretch to bring this to earthquakes, but that's what I do. I'm glad they are doing this in the lab, since I was looking at this for earthquakes 30 years ago. Non-linearity simply means they blast the hell out of it. With dynamics, all materials have a low stress linear zone, and then it starts to curve as the amplitude gets higher. With imaging, they have to be careful not to go too high.
This just dips into the fine cracks. With earthquake seismic waves, the linear zone starts at about 10 km away. In the East, when the seismic waves hit this sweet spot, they can propagate 'forever'. But nothing gets damaged in this zone, unless it is a brick building on a deep swamp. This is what those 'early warning systems' want to protect us from. When I did a study for a deep nuclear repository, I found these seismic waves are limited in their induced stress to a very small level. That is why the PGV is so low on rock.
In the immediate fault zone, all bets are off. Non-linearity is in full swing. The PGV can really get up there. This is also the reason I went for PGV exclusively, it directly reflects the induced stress of the wave.