Is deep non-volcanic tremor just a swarm of low frequency earthquakes?

It is commonly assumed that tremor is a superposition of many low frequency earthquakes (LFE) [Shelly, D.R., Beroza, G.C. & Ide, S., 2007, Nature, 446, 305]. In the model we propose, tremor is a result of specific radiation generated inside a fault (Figure 1). This radiation may be initiated by LFE as well as by other types of local failure, such as regular earthquakes, very low frequency (VLF) earthquakes and aseismic slip. The specific features predicted by the model are: (1) the central frequency of tremor is defined by the effective normal stress at the source location; (2) a comparatively small effective normal stress (hence a high fluid pressure) is required to make the model consistent with observed tremor parameters; (3) the size of the emitting area is on the order of a few dozen km; (4) the frequency content of tremor depends on the type of initial failure. "Fast" sources, such as regular earthquakes and LFE, produce larger amplitudes at high frequencies than "slow" sources such as VLF earthquakes and aseismic slip. Figure 1. Scenario of tremor generation. The initial failure (black object at the bottom panel), triggered by a seismic wave from a distant earthquake, produces oscillations (shown by red arrows) propagating along a subduction fault (in yellow). These oscillations generate S-type seismic waves (tremor) propagating to the surface. The graphs show velocity v(t) at the Earth's surface (top) and relative shift u(x) inside a fault (bottom).