I'm trying to model propagation of these waves from oceanic sources. Webb (Nature 2007) suggests a low Q for periods between 20 and 40 s, or something like this. Is there any new knowledge on this issue of Q factors?
From 10 to 20 s, I guess attenuation due to scattering is dominant, because of large lateral variations of thickness of the water layer. In this frequency range, attenuation due to the sedimentary layer is also important. A recent paper of "FDM simulation of an Anomalous Later Phase from the Japan Trench Subduction Zone Earthquakes" by Noguchi et al. (2013) may be informative.